Geophysical Monitoring of Moisture-Induced Landslides:A ReviewJ. S. Whiteley1,2 , J. E. Chambers1 , S. Uhlemann3 , P. B. Wilkinson1 , and J. M. Kendall2

1Environmental Science Centre, British Geological Survey, Nottingham, UK, 2School of Earth Sciences, University of Bristol,Bristol, UK, 3Formerly British Geological Survey, now Lawrence Berkeley National Laboratory, Berkeley, California, USA

Abstract Geophysical monitoring of landslides can provide insights into spatial and temporal variations ofsubsurface properties associated with slope failure. Recent improvements in equipment, data analysis, andfield operations have led to a significant increase in the use of such techniques in monitoring. Geophysicalmethods complement intrusive approaches, which sample only a very small proportion of the subsurface,and walk-over or remotely sensed data, which principally provide information only at the ground surface. Inparticular, recent studies show that advances in geophysical instrumentation, data processing, modeling, andinterpretation in the context of landslide monitoring are significantly improving the characterization ofhillslope hydrology and soil and rock hydrology and strength and their dynamics over time. This reviewappraises the state of the art of geophysical monitoring, as applied to moisture-induced landslides. Here wefocus on technical and practical uses of time-lapse methods in geophysics applied to monitoringmoisture-induced landslide. The case studies identified in this review show that several geophysicaltechniques are currently used in the monitoring of subsurface landslide processes. These geophysicalcontributions to monitoring and predicting the evolution of landslide processes are currently underrealized.Hence, the further integration of multiple-parametric and geotechnically coupled geophysical monitoringsystems has considerable potential. The complementary nature of certain methods to map the distribution ofsubsurface moisture and elastic moduli will greatly increase the predictive and monitoring capacity of earlywarning systems in moisture-induced landslide settings.

1. Introduction

The destabilization and subsequent mass movement of soil and rock on slopes occurs across the globe, lead-ing to loss of life and damage to property and infrastructure (Froude & Petley, 2018; Petley, 2012).Investigation of landslides can determine their key characteristics, including (but not limited to) soil and rockproperties, the landslide geomorphology, types of movement, and velocity rates. Detailed knowledge ofthese characteristics can inform the modeling of the sensitivity of landslide masses to external triggers(Arnone et al., 2011; Jibson, 1993), which contribute to reducing risk posed by these globallyoccurring hazards.

The worldwide distribution of landslides is not uniform, with landslides occurring primarily where the requi-site topographic, climatic, and environmental conditions are prevalent. Figure 1 shows the distribution ofmoisture-induced landslides (i.e., those caused by increased hydrological infiltration) recorded in theGlobal Landslide Catalog between 2007 and 2016 (Kirschbaum et al., 2010, 2015). An obvious pattern isthe greater occurrence of landslides in areas of greater topographic variation compared to areas of relativelylow relief.

High-incidence, high-fatality areas tend to be found in less developed regions (e.g., South East Asia, Centraland South America) and high-incidence, low-fatality areas are generally located in more developed countries(e.g., North America, Europe). However, the Global Landslide Catalog does not capture the whole picture oflandslide distribution and impact across the globe (Kirschbaum et al., 2015). In the United Kingdom, there is awell-established and maintained recording system for landslide events, operated by the British GeologicalSurvey (Pennington et al., 2015). The United States Geological Survey has a national landslidemitigation strat-egy which recommends the mapping and assessment of landslide hazards (Spiker & Gori, 2003). These land-slide recording programs tend to capture the occurrence of both fatal and nonfatal landslide events at manyscales. However, in many other countries, particularly developing nations, such recording systems are notestablished, and the landslide events recorded tend to be those that have a large enough impact on life,

WHITELEY ET AL. 1

Reviews of Geophysics

REVIEW ARTICLE10.1029/2018RG000603

Key Points:• Geophysical monitoring reveals high

resolution spatial and temporalinformation from the subsurface oflandslide systems

• Developments in geophysicalmonitoring have led to substantialincreases in applications to landslidesand frequencies of data acquisition

• Linking geophysical data withgeotechnical measurements canmonitor hydrological andgeomechanical processes in timeand space

Correspondence to:J. S. Whiteley,jim.whiteley@bristol.ac.uk

Citation:Whiteley, J. S., Chambers, J. E.,Uhlemann, S., Wilkinson, P. B., & Kendall,J. M. (2019). Geophysical monitoring ofmoisture-induced landslides: A review.Reviews of Geophysics, 57. https://doi.org/10.1029/2018RG000603

Received 6 APR 2018Accepted 6 NOV 2018Accepted article online 15 NOV 2018

©2018. The Authors.This is an open access article under theterms of the Creative CommonsAttribution License, which permits use,distribution and reproduction in anymedium, provided the original work isproperly cited.

society, or the economy to be reported in the media. Although fatalities are fewer in more economicallydeveloped countries, there are still issues of completeness in reporting landslides even in Europe (Haqueet al., 2016). The incidence of landslide events across the globe can therefore be taken as a lower boundaryof actual landslide occurrence, and loss of life and infrastructure from landslides can be greatly underesti-mated in such circumstances (Petley, 2012). Landslides are often considered a secondary hazard, formingpart of the problem of cascading hazards (Pescaroli & Alexander, 2015), as they are triggered by catastrophicevents such as storms, floods, volcanic eruptions, and earthquakes (Gill & Malamud, 2014). This cascadingeffect was highlighted in Papua New Guinea in February 2018, with numerous landslides triggered by a largeearthquake blocking many valleys. This valley blocking poses a continuing flood risk as water builds upbehind these unstable structures with increased seasonal rainfall.

The term “moisture-induced landslides” (MIL) is used in this review to refer to landslides triggered byincreased water infiltration. Elevations in subsurface moisture content can be caused by extended periodsof increased infiltration, or by increased volumes of water. Most infiltration in landslide settings comes fromincreased precipitation or snowmelt, and due to the complex role of evapotranspiration, the amount of rain-fall or snowmelt that reaches the subsurface can vary throughout the climatic cycle. The amount of moisturethat does reach the subsurface is known as “effective infiltration.” Therefore, the termMIL recognizes themul-tiple sources of moisture that affect landslides, the importance of the concept of effective infiltration, and itssubsequent effect on the instability of materials in response to moisture infiltration.

The advances in hardware and software for geophysical monitoring reflect those made in remote sensortechnology, mainly in the ability to deploy increasingly low-cost and low-power sensors to capture informa-tion from the subsurface with a minimal delay in data acquisition and transmission (Ramesh, 2014). The mainbenefit arising from these developments has been the increase in monitoring durations achieved by theinstallation of permanent sensors (Ramesh & Rangan, 2014). However, geophysical monitoring approacheshave the added benefit of providing increasingly high-resolution spatial information.

This review appraises the state of the art of the application of geophysical methods to monitoring moisture-induced landslides. Geophysical monitoring provides information on fundamental subsurface slope process,and is relevant to those studying landslides and their behavior. However, this information can also be vital tothose looking to mitigate landslide hazards, and in particular, can provide information on precursory failureconditions in unstable slopes. Therefore, the content of this review is also of interest to early warning systems(EWS) operators looking to incorporate spatial and temporal subsurface data into their monitoring systems.

Figure 1. Global occurrences of landslides recorded in the global landslide catalog between 2007 and 2016 (Kirschbaumet al., 2010; Kirschbaum et al., 2015), including associated fatalities. Redrawn from Uhlemann (2018).

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 2

In this review, “geophysical methods” specifically refer to surface deployed techniques to investigate featuresin the shallow subsurface. The term “monitoring” is used to indicate a time-lapse approach to investigatechanges between two or more geophysical data sets acquired at the same location at different times.Recent literature shows two main methods being employed for the monitoring of landslides, geoelectricaland seismic. The relevance of geophysical monitoring to landslides is addressed by first identifying landslidecharacteristics and the application of geophysical methods. A review of case studies of geophysical monitor-ing of landslides is then undertaken in light of the methods currently in use. Finally, recommendations for theuse of geophysical methods in themonitoring of landslides will be revisited in the context of slope-scale EWS.

2. Landslide Settings and Processes: Definitions

The most longstanding and recognized classification of landslides is that of Cruden and Varnes (1996). Thisclassification system has been updated in recent years to reflect modern standards of material properties(Hungr et al., 2014). Other well-established and regularly used systems exist for other specific landslide types(e.g., Fell, 1994; Leroueil et al., 1996; Skempton & Hutchinson, 1969). The updated Cruden and Varnes (1996)system by Hungr et al. (2014) comprehensively outlines the types of movement, the geological materials, andvelocities of material movement that describe the majority of landslide occurrences across the globe.

For this review, it is useful to distinguish between the spatial definition of a landslide setting, and the tem-poral description of the evolving processes of movement:

1. The “landslide setting” is the spatial definition of a geographical area that may be prone to, or have experi-enced, the downslope mass movement of geological material (e.g., Jongmans et al., 2009). Characteristicsof the landslide setting would include the geological materials (rock, boulders, debris, sand, clay, silt, mud,peat, ice) as well as the geomorphology of the unstable slope (Guzzetti et al., 1999; Hungr et al., 2014).

2. “Landslide processes” indicate the onset and subsequent processes of movement of unstable geologicalmaterial in time (e.g., Hutchinson & Bhandari, 1971). These include the changes in the subsurface condi-tions of landslides preceding observable failure, such as elevations in pore water pressure that may inducefuture movement in landslides. These processes may be difficult to measure, and often can only beinferred by observing changes in a property (e.g., groundmoisture) over time. Factors relating to landslideprocesses would include the types of movement (flows, topples, slides, spreads, and deformations), velo-cities of movement (<16mm/year to>5m/s; Hungr et al., 2014), and the hydrogeological and geomecha-nical processes acting upon the landslide materials.

Identification of the affected area is typically undertaken by walk-over surveys, studying of aerial photo-graphs, or via other remote sensing methods such as satellite imagery (Carrara et al., 1992, 2003; Guzzettiet al., 2012). Landslide investigations using geotechnical means typically involve identification of the three-dimensional subsurface structure of the landslide, the definition of the hydrogeological regime, and thedetection of movement in the landslide mass (Mccann & Forster, 1990). This has traditionally been underta-ken with intrusive investigation methods, such as the use of trial pits and boreholes to recover samples andidentify changes in materials and their properties within the body of a landslide (Angeli et al., 2000;Uhlemann, Smith, et al., 2016). These approaches allow for a great amount of detailed information to be gath-ered at discrete locations. However, due to the heterogeneous subsurface conditions of landslide settingsand the dynamic response of landslide processes to external influences, they are not always adequate in pro-viding a wider view of the landslide system, both spatially and temporally.

3. Landslide Geophysics: An Overview

Investigations of subsurface landslide features are necessary to provide the input for forward modeling andsubsequent predictions of potential failure events, for example, estimating the runout length, the mobilizedvolume, or the velocity of a potential failure event (Malet et al., 2005; Rosso et al., 2006). In general, geophy-sical techniques identify spatial variations of a physical parameter of the subsurface, from which inferenceson a range of processes and properties can be made (Everett, 2013; Kearey et al., 2001; Parsekian et al.,2015). When applied to landslide investigation, geophysical techniques are able to target characteristicsand features of landslide settings that are manifested by physical property contrasts in the subsurface(Mccann & Forster, 1990), including the following:

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 3

1. The physical extent of the landslide, comprising critical features such as the subterranean slip surface andwater table

2. Variations in lithological and soil units3. Variations in distribution and movement of moisture throughout the landslide body4. Variations in the geomechanical strength of the landslide body

3.1. The Landslide Setting and Geophysical Investigation

The features of a typical landslide system that can be identified and assessed using geophysical methods areshown in Figure 2. These features are typified by the existence of a physical discontinuity (e.g., slip surface,lithological contact) or a contrast in material properties (e.g., degree of saturation, clay content) being presentin the subsurface. Table 1 summarizes the main landslide features identified in Figure 2, and lists the targeteddiscontinuity or property contrast, and the applicable geophysical techniques for identifying these features.

3.2. Landslide Processes, Soil Mechanics, and Geophysical Investigation

The major parameters influencing slope stability are summarized in Figure 3. This summarizes the classicalunderstanding of slope failure through the principles of soil mechanics (Terzaghi, 1943; Terzaghi et al.,1996), indicating the key geotechnical parameters acting upon unstable slopes. The figure is a simplifiedmodel of a slope, making many assumptions, such as the length of the slope, but indicates the main propertyto consider when estimating slope stability. The Mohr-Coulomb failure criterion (Terzaghi, 1943) determinesthe shear strength (τf) of the material at a given point at the slip surface interface, and is given by

τf ¼ c þ σ � uð Þ tanϕ 0cv ; (1)

where c is the cohesion, σ is the total normal stress, u is the pore water pressure, and ϕ0cv is the angle of shear

resistance at a critical state.

The total normal stress (σ) is given by

Figure 2. Schematic of a landslide system, showing the major landslide setting features that can be investigated andassessed using geophysical methods.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 4

σ ¼ mγþmγsatf gz cos2β; (2)

where m is the height, γ is the unit weight of material, γsat is the saturated unit weight of material, z is thedepth to slip surface, and β is the slope angle. Shear stress (τ) is given by

τ ¼ mγþmγsatf gz sinβ cosβ: (3)

The pore water pressure (u) at the slip surface is calculated by

u ¼ mzγw cos2β: (4)

In order for slope failure to occur, the restraining forces, in this case shear strength (τf), must be overcome bydisturbing forces, or shear stress (τ). This can be expressed in terms of a factor of safety (FoS), given by

Table 1Summary of the Landslide Features Able to be Identified, Assessed, and Investigated Using Geophysical Methods

Feature of theLandslide Setting Discontinuity or Property Contrast Applicable Geophysical Methods Example

Landslide extents(x, y, z)

Slip surface (subsurface and surficial extent)caused by or indicated by changes in density,water content, etc., of material

Electrical resistivity, seismic reflection, seismicrefraction, surface wave methods, ground-penetrating radar

Chambers et al. (2011)

Subsurface materialtype and structure

Material density Microgravity Sastry and Mondal (2013)Relative degree of saturation Electrical resistivity, electromagnetics Springman et al. (2013)Relative clay content Electrical resistivity, electromagnetics Göktürkler et al. (2008)Material velocity (as a function of density) Seismic reflection, seismic refraction,

surface wave methodsRenalier, Jongmans, et al. (2010)

Water table Height of water table Electrical resistivity, electromagnetics,seismic refraction

Le Roux et al. (2011)

Relative flow direction Self-potential Perrone et al. (2004)Tension features(e.g., surface fractures)

Saturation contrasts (e.g., preferentialinfiltration pathways)

Electrical resistivity, electromagnetics,ground-penetrating radar, self-potential

Bièvre et al. (2012)

Compression features(e.g., ridges)

Variations in material composition Electrical resistivity, electromagnetics,ground-penetrating radar

Schrott and Sass (2008)

Figure 3. Schematic of an infinite slope model showing the main landslide processes showing the main components of aclassical soil mechanics approach to slope failure (Terzaghi, 1943), redrawn from Craig (2004).

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 5

FoS ¼ τfτ: (5)

When FoS< 1 slope failure will occur, with FoS> 1 indicating stable slope conditions. FoS< 1 can be reachedby increasing τ, for example, by greater loading on the slope, or by reduction in τf, for example, by increasingu (equation (1)). Not highlighted in these equations is the critical role that negative pore water pressures, orsoil suction (or matric suction) can play in the stabilization of landslide bodies (Toll et al., 2011). Materials withlarger void spaces, such as sands and gravels, will have smaller capillary zones (the area of saturation abovethe water table caused by soil suction) than cohesive materials with smaller void spaces such as clays and silts(Craig, 2004). In some slopes, soil suctions may be the main restraining force preventing failure (Hen-Joneset al., 2017). In these settings, understanding of subsurface moisture dynamics relating to increased infiltra-tion is critical for predicting future slope failures.

In a geophysical survey, the result may determine the presence or extents of a landslide property (e.g., thewater table). This time-static geophysical data set gives information on the state of one (or more) propertyin the system at that particular time, but not how that property will change in response to some externalinfluence (e.g., precipitation). With the addition of a second geophysical data set at some point in the future,it may be possible to determine a change in that property (e.g., increase in water table height). The implicitassumption is that in order for a property change to have occurred, a process must have taken place in thetime between the two surveys (e.g., infiltration). By looking at the differences between the data at each end ofthe time period, reasonable inferences can bemade about the process that must have occurred to give rise toa change in the system.

Therefore, geophysical methods used in isolation are not able to definitively quantify the contributions thatchanges in these properties make to the overall stability of a slope, but the incorporation of environmentaldata (e.g., rainfall data) and the use of geotechnically coupled surveys can provide empirical relationshipsthat allow for such contributions to be assessed. Table 2 shows the parameters of landslide processes towhich various geophysical methods are sensitive.

The figures presented in this section are simplified versions of landslide systems, meant to illustrate the fea-tures on landslide settings and processes that can be investigated in some way using geophysical methods.Field conditions are likely to present more heterogeneous conditions than those represented here. However,the identification of such heterogeneous conditions is itself an advantage of using geophysical methods toinvestigate landslide, with the spatial coverage of geophysical methods being one of the main benefits ofthe techniques (Everett, 2013; Kearey et al., 2001).

3.3. Applications of Geophysical Methods to Landslide Investigation

For a geophysical methodology to be effective as a monitoring tool, several criteria must be fulfilled:

Table 2Parameters of Landslide Processes That Can Be Investigated Using Geophysical Methods

Slope Stability Property Investigable Feature Applicable Geophysical Methods Example

Normal stress (σ)σ = {mγ + mγsat}zcos

2βDepth to slip surface (z) Electrical resistivity, seismic reflection,

seismic refraction, surface wavemethods, ground-penetrating radar

Renalier, Jongmans, et al. (2010)

Relative proportions of dry (or partiallysaturated) material ((1 � m)γ)

Electrical resistivity, seismic reflection,seismic refraction, ground-penetrating radar

Grandjean et al. (2010)

Relative proportions of saturatedmaterial (mγsat)

Pore water pressure(u)u = mzγwcos

2βHeight of water table (mz) Electrical resistivity, seismic reflection,

seismic refraction, ground-penetrating radarSastry and Mondal (2013) and Grelleand Guadagno (2009)

Cohesion (c) Presence of clay in subsurface material Electrical resistivity, ground-penetratingradar

Göktürkler et al. (2008)

Shear strength (τf)

τf¼cþ σ-uð Þ tanϕ0cv

Estimates of angle of shear resistance atcritical state (φ’cv) can be made using

back analysis from estimating σ and u

Seismic refraction, surface wave methods Al-Saigh and Al-Dabbagh (2010)

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 6

1. The subsurface property or process (e.g., moisture content) beingmeasured or monitoredmust be detect-able by the chosen method.

2. The acquisition rate of data must be sufficiently high (with respect to changes in the monitored process)so as to be able to collect data sets which sample physical property changes at an appropriate rate.

3. The measuring equipment must be able to be relocated accurately for series of discrete fieldwork cam-paigns, or the position of geophysical sensors must be accurately located if a semipermanent installationis established.

4. The data must ideally be able to be empirically linked to subsurface conditions so as to confidently reflectthe changes that occur in the physical properties of the subsurface over time.

In recent years, the application of geophysical methods to the characterization of landslide settings hasbecome increasingly common. Hack (2000) outlined the physics underpinning geophysical surveys for slopestability analyses, and Jongmans and Garambois (2007) provided a review of published landslide investiga-tions using geophysical methods since the mid-1990s.

Both of these reviews highlight three important developments in landslide geophysics: the integration ofmultiple methods, the acquisition of multidimension data (including time-lapse approaches), and the quan-tification of geophysical results with geotechnical data. The integration of multiple geophysical methods forlandslide investigations has become more common in recent years, allowing for better evaluation of internalstructure to be made by overcoming the limitations of single-technique approaches (Schrott & Sass, 2008).However, multiple-method approaches to landslide monitoring using geophysical methods are still relativelyscarce (Jongmans & Garambois, 2007).

The spatial dimensions across which geophysical data can be acquired have increased over time. One-dimensional (1-D) “sounding” type data provides information on the subsurface beneath a single locationon the ground surface. These vertical profiles of information can be interpolated to form pseudo-two-dimensional and pseudo-three-dimensional data sets of subsurface properties.

True two-dimensional (2-D) survey techniques allow data to be collected spatially across the ground surface(also referred to as “geophysical mapping”), or as a subsurface profile or cross section. The latter of these tech-niques can be interpolated to produce pseudo-three-dimensional volumes of data. True three-dimensional(3-D) surveys involve the simultaneous acquisition of volumes of data. All of these dimensions of data acqui-sition produce a static map of the distribution of physical properties within a landslide body, and proveinvaluable for the characterization of landslide settings. However, in order to work toward the predictionof landslide failure, an investigation method that takes into account the temporal variation within the land-slide system should be used.

Geophysical techniques that include the addition of a time series to the acquired data are sometimes referredto as “time-lapse” or “4-D” data sets. As 4-D specifically refers to a 3-D data set with multiple time steps, theterm time lapse will be used in this review to refer to data sets of any dimension (1-D, 2-D, and 3-D) whichinclude multiple time steps. Time-lapse geophysical monitoring of landslides is an area that has grownrapidly in the last decade or so (Jongmans & Garambois, 2007).

In order for the application of geophysical techniques to the monitoring of MIL over time to be successful, thefrequency of data set acquisition must be proportional to the time scale over which changes occur in thesubsurface properties of the landslide. Geophysical monitoring therefore lends itself to the monitoring ofslowly deforming slopes. These slopes are typically composed of highly weathered rocks or soils and othersuperficial materials (such as tills, alluvium, and other recently reworked materials). Failure in these landslideenvironments can be both “brittle” (i.e., failure occurs along discreet shear surfaces) or “ductile” (i.e., thedeformation processes are slow, such as slope creep). Landslides triggered by increased infiltration of waterare of particular interest to geophysicists, as the formation and progression of precursory failure conditionscan be detected using geophysical data, processing, and interpretation (Baroň & Supper, 2013). For this rea-son, fast-failing and fast-moving landslides that do not typically exhibit gradual changes in subsurface prop-erties prior to failure have been excluded from this review. These landslide types are typically fast-failingrockfalls and rock topples, fast-moving debris flows, and fast-failing and fast-moving rock avalanches(Hungr et al., 2014). In some instances, these types of failure may also be present in a MIL setting (e.g.,Helmstetter & Garambois, 2010; Lacroix & Helmstetter, 2011) It is important to note, however, that geophy-sical monitoring systems do exist for these types of failure (Fiorucci et al., 2016; Lotti et al., 2015;

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 7

Partsinevelos et al., 2016) and that many of the geophysical methods described here are similar in theirapplication to these failure types. The case studies presented in this review therefore have a focus on themonitoring of landslides in settings comprising of soft-rock and superficial materials.

MIL are typically shallow, translational, and/or rotational style landslides, with failure frequently occurring insoils and weathered bedrock layers. An exception to this is deep-seated gravitational slope deformation-typefailures, which are characterized by very slow deformations of very large rock masses (Jomard et al., 2010;Lebourg et al., 2005). These have been included in this review as the progression of failure is slow enoughto be robustly monitored using geophysical approaches, and periods of increased movement are linked toincreases in precipitation (Helmstetter & Garambois, 2010; Lacroix & Helmstetter, 2011; Palis, Lebourg,Vidal, et al., 2017). The study of landslides and their evolving processes is a highly interdisciplinary area ofresearch. As such, the application of monitoring MIL has important implications for geologists, geomorphol-ogists, geotechnical engineers, and infrastructure stakeholders, especially as the impact of climate change ishaving a dramatic effect on the incidence of slope failures due to changing weather patterns (Gariano &Guzzetti, 2016).

4. Geophysical Monitoring of Landslides: Methods and Case Studies

A list of publications in scientific journals and books since 2006 that utilize geophysical monitoring applied tolandslides is shown in Appendix A. In total, 38 journal articles or book chapters are identified as using ordescribing geophysical monitoring approaches in landslide systems. Of these 38 publications, two are projectupdates that do not contain data, and 36 describe specific case studies. Some of the individual monitoringcampaigns are described in more than one publication, and where possible this is indicated, leaving a totalof 34 publications. Within these 34 publications, 54 monitoring campaigns at 27 separate landslides are iden-tified from nine countries.

The geophysical monitoring methods identified and their corresponding abbreviations for this paper areshown in Figure 4. Further descriptions of these methods are found in the relevant sections detailing the casestudies. Two broad types of geophysical methods are identified from this list: geoelectrical and seismic meth-ods. In addition, two modes of acquisition are identified in both types: “active”modes, in which the recordedgeophysical signal is artificially generated, and “passive” modes, in which the signal recorded isgenerated naturally.

In geoelectrical methods one active type, electrical resistivity (ER), and one passive type, self-potential (SP)monitoring, are identified. Typically, 2-D electrical resistivity (ER) profiles are acquired as part of the study,with similar instrumentation, acquisition parameters, processing routines, and inversion methods appliedacross the case studies. Some examples where geoelectrical data sets are manipulated in bespoke ways doexist, for example, the use of apparent resistivity (Palis, Lebourg, Tric, et al., 2017), transfer resistance data(Merritt et al., 2018), and 3-D ER arrays (Uhlemann et al., 2017). The use of SP methods is also noted, althoughin a much reduced capacity to ER (Colangelo et al., 2006). In total, 20 time-lapse ER monitoring surveys are

Figure 4. The geophysical methods identified in the 31 case studies, shown by mode of acquisition and method. The acro-nyms for each method are shown in the lower boxes.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 8

identified in the literature: Bièvre et al. (2012), Crawford and Bryson (2018), Friedel et al. (2006), Gance et al.(2016), Grandjean et al. (2009), Jomard et al. (2007), Lebourg et al. (2010), Lehmann et al. (2013), Lucaset al. (2017), Luongo et al. (2012), Merritt et al. (2018), Palis, Lebourg, Vidal, et al. (2017), Palis, Lebourg, Tric,et al. (2017), Supper et al. (2014), Travelletti et al. (2012), Uhlemann et al. (2017), and Xu et al. (2016). Onetime-lapse SP monitoring survey is identified, by Colangelo et al. (2006).

In seismic methods, the active types of surveying include seismic refraction (SR) and analysis of surface waves(SW) and the passive types include seismic event detection, characterization, and location (S-EDCL); seismiccross correlation (S-CC); seismic horizontal-to-vertical ratio (S-H/V); and seismic ambient noise tomography(S-ANT). One study by Grandjean et al. (2009) used SR monitoring, and a study by Bièvre et al. (2012) utilizedSW monitoring methods. In total, 31 case studies are identified as using passive seismic methods: Amitranoet al. (2007), Brückl and Mertl (2006), Brückl et al. (2013), Gomberg et al. (2011), Harba and Pilecki (2017),Imposa et al. (2017), Mainsant et al. (2012), Palis, Lebourg, Vidal, et al. (2017), Provost et al. (2017), Renalier,Jongmans, et al. (2010), Tonnellier et al. (2013), Walter and Joswig (2008), Walter et al. (2009), Walter et al.(2011), Walter et al. (2012), and Walter et al. (2013).

Some geophysical methods are absent from the literature surrounding landslide monitoring; for example, nostudies in which ground penetrating radar has been used as a long-term monitoring tool are identified,although it has been used to provide detailed qualitative characterization of landslide settings (Hruska &Hubatka, 2000).

Before presenting the details of individual methods and case studies, it is helpful to consider the differentconditions under which the geophysical monitoring campaigns listed in Appendix A were acquired. Themostimportant aspects of these conditions to consider are the duration of monitoring, the modes of acquisition,and the frequency at which data were acquired during the campaign.

4.1. Duration of Geophysical Monitoring Data Acquisition

The durations of geophysical monitoring were able to be divided in to four approaches:

1. Controlled tests: Typically artificial infiltration-type experiments in which an unstable slope was exposed toan increase in saturation by introducing simulated rainfall over a set period of time and subsurface con-ditions monitored throughout. Although the field instrumentation remained static between acquisitions,the field setups were not typically designed for long-term monitoring, and were assumed to have beenaccompanied by field crew for the duration of the experiment. The shortest and longest controlled experi-ments were both described by Lehmann et al. (2013), at 15 hr in one test and three days in another. Allcontrolled experiment case studies showed a high-frequency of data set acquisitions, typically everyfew hours. Six controlled tests were described in five publications: Grandjean et al. (2009), Jomard et al.(2007), Lehmann et al. (2013), Travelletti et al. (2012), and Colangelo et al. (2006).

2. Transient measurements: Typically involved the deployment of instruments for the duration of a singledata set acquisition, after which the instrumentation was removed from the field before being deployedagain in the same location after a period of time. Monitoring periods varied greatly, with the shortestrecorded as 62 days (Lucas et al., 2017) and the longest as 1589 days (Imposa et al., 2017). In both of thesestudies, six data sets were acquired over the monitoring period, highlighting the variability in acquisitionfrequency that can accompany a transient measurement approach.

3. Short-term installations: Utilize equipment installed and left in the field, but for shorter periods than semi-permanent installations. Consequentially, the length of these monitoring campaigns fall between con-trolled tests and semipermanent installations. Unlike controlled tests, environmental conditions are notartificially altered (e.g., by artificial rainfall), and instead, natural variations in environment and subsurfaceresponse are monitored. However, equipment tends to be deployed for periods of less than 100 days, andso surveys cannot typically monitor seasonal- or annual-scale variation subsurface conditions, althoughthe monitoring period may coincide with periods of increased rainfall. Often, the full cycle of a monitoringcampaign may involve several short-term installations at the same landslide (e.g., Brückl & Mertl, 2006),although in some studies, the exact position can vary between these repeated surveys (e.g., Walteret al., 2011). The shortest period of monitoring using short-term installations was for one day, althoughthis day comprised one survey of a sequence described by Brückl and Mertl (2006). The longest periodof monitoring achieved with a short-term installation was by Lebourg et al. (2010) at 90 days.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 9

4. Semipermanent installations: Installations are left in the field to acquire data according to a predetermined(or sometimes remotely programmable) schedule. Unlike transient measurements, the time periodbetween acquisitions was often regular. Semipermanent installations allow for a high-frequency dataacquisition (typically one or more data sets acquired per day) over long monitoring periods (typicallymonths to years). The shortest monitoring period studied with a semipermanent installation was 146 daysin a study by Luongo et al. (2012). Most studies operating at annual-scale time periods, and often at anacquisition rate of more than one data set per day. In Supper et al. (2014), two examples are given, wheredata sets were recorded every 4 hr for 239 and 275 days at two separate sites. These monitoring systemsalso tend to include the ability for data to be retrieved remotely from the equipment, often by utilizingtelemetry and storage of data on a remote server.

4.2. Acquisition Mode of Geophysical Monitoring Data

Figure 5 shows the relationship between different geophysical methods, and how they are suited to differenttypes of geophysical monitoring. Geophysical methods acquiring active mode data lend themselves toacquiring higher-resolution spatial data due to the tendency toward acquiring time-discreet data sets.Geophysical methods acquiring passive mode data tend to higher-resolution temporal data, as they acquiretemporally continuous data sets. Some geophysical methods, however, are able to produce high-resolutionspatial data from passive acquisition modes.

4.3. Geophysical Data Acquisition Frequency

Figure 6 shows a chart plotting the duration of each monitoring campaign against the number of data setsacquired in each monitoring campaign. There is some difficulty equating the number of data sets acquired

Figure 5. Relationship between the geophysical methods outlined in Appendix A in terms of their acquisitionmode (activeor passive) and temporal resolution (continual or time-discreet) demonstrating the trade-offs which must be consideredwhen choosing a method for monitoring of MIL.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 10

in a monitoring period between different types of geophysical methods. The ER, SP, SR, SW, S-ANT, and someS-H/V methods acquire time-discreet sets of data. This is because the entirety of a data set is made up of amultitude of unique data points, the position and number of which determine the spatial resolution of thesurvey method. These methods therefore tend to have high spatial resolution and coverage (see Figure 5).It is the number of electrodes (in ER and SP surveys), geophones (in SR and SW surveys), or seismic sensors(in S-ANT and S-H/V surveys) that define the spatial resolution and coverage; the more measuring pointsto the system, the higher the resolution and/or the larger the spatial coverage of the monitoring system.However, the temporal resolution is limited to the amount of time it takes to acquire the full complementof unique readings in a whole data set. In ER systems, for example, it can take several hours for a full dataset to be acquired.

In contrast, S-EDCL, S-CC, and some S-H/V methods acquire data from continually recording seismometers.In these cases, data are acquired for the entirety of the monitoring period, with the exception of any per-iods of maintenance or failure. These methods therefore have high temporal resolution (see Figure 5);however, similarly to discreet data set acquisition systems, the spatial resolution is still dependent onthe number of measuring components. For some applications (e.g., S-CC and S-H/V), data are divided intodaily records, but in some methods (e.g., S-EDCL), only the time of the event recorded is required. Thenumber of data sets acquired in a monitoring period by seismometers is therefore technically one dataset per field campaign, although this can be completely arbitrary in some processing methods. In someapproaches, such as in S-ANT, the distinction between continuously acquired and time-discreet data isblurred further, as the data processing typically deals with discrete time steps extracted from the contin-uous data set.

Therefore, in Figure 6, for time-discreet data set acquisition systems (ER, SP, SR, SW, S-ANT, and some S-H/Vmethods), the number of data sets acquired across the entire monitoring period are plotted. For monitoringcampaigns that utilize continually recording seismometers (S-EDCL, S-CC, and some S-H/V methods), thenumber of seismometers, rather than the number of data sets acquired, has been plotted. For seismometerentries, this therefore gives an indication of the spatial resolution of the survey, rather than the temporal reso-lution, which can be assumed to be total for the monitoring period.

Figure 6. Plot showing the year of publication for case studies against the number of days in the monitoring period foreach case study. The relative area of each data point is proportional to the number of data sets acquired. Data collectedfrom seismic monitoring networks are considered to have collected one data set per sensor for the entire monitoringperiod.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 11

With regards to the spatial resolution and coverage of surveys, an ER system for example, may have a singleunit (with a single data logger, single power source, single telemetric communication system) controlling anentire electrode array consisting of several tens of electrodes, across a profile of hundreds of meters.Conversely, a single seismometer, or localized array of seismometers, is typically operated by a single loggerwith a single power source and the coverage of data acquisition is dependent on the events detected at thatlocation. To expand the spatial resolution, a second seismometer with a separate logger and power sourcemust be added. In contrast, expanding the spatial coverage of an ER monitoring system may be as simpleas adding extra electrodes in to the system.

Figure 6 illustrates several interesting developments in the field of geophysical monitoring of MIL in the lastdecade or so. First, the number of published studies is increasing in time, suggesting increases in the applica-tion of geophysics to landslide monitoring, or in some cases, the recent maturation of long-term studies.Second, there is a notable increase in ER case studies from 2012, and a sharp increase in both the monitoringperiod lengths and amount of data collected by these studies. This suggests greater developments in thefield of monitoring landslides using ER methods, compared to seismic methods which show much morerestrained increases. This trend is curious, as passive seismic instrumentation has been at a far moreadvanced state of development for long-term monitoring for much longer than ER methods.

4.4. Geoelectrical Monitoring Case Studies

Moisture dynamics can vary greatly over time, although lithological composition tends to remain relativelyconstant, particularly in a prefailure condition. Geoelectrical monitoring can therefore be used for the deter-mination of moisture dynamics and hydrogeological processes within MIL bodies over time.4.4.1. Active Geoelectrical Methods4.4.1.1. ERElectrical resistivity (ER) is a common technique routinely applied to the investigation of landslides(Jongmans & Garambois, 2007). The method can be applied in 1-D, 2-D, and 3-D investigations (Loke et al.,2013), although 2-D and 3-D investigations of landslides are more common in recent times. ER is measuredby injecting a DC current into the ground between two electrodes, and measuring the potential differencebetween a separate pair of electrodes (Kearey et al., 2001). By deploying linear arrays of electrodes, measure-ments can be made using different combinations of electrodes in order to build a 2-D image of the subsur-face resistivity. Similarly, 3-D data sets can be acquired using multiple parallel (and orthogonal) profiles orgrids of electrodes.

In a typical 2-D linear array, the number of uniquemeasurements made is proportional to the number of elec-trodes deployed at the ground surface. The position of a single resistivity measurement is a function of thedistance between the electrodes used, and their position on the ground surface. In most near-surfacegeophysical applications, the raw data recorded by ER meters (“apparent resistivity”) is processed using atomographic inversion. The process of tomographic inversion produces a ground model based on the datarecorded and a set of assumptions made from physical laws (e.g., current flow through subsurface equipoten-tial surfaces) and Earth observations (e.g., subsurface models obeying geological principles). Crucially, theprocess provides information between the measurement points of an array based on the surrounding data.Inversions can be done in 2-D and 3-D, and with the addition of time-lapse sequences in which the time seriesbetween data set acquisitions is treated as a variable in the inversion process. However, a common approachis to undertake separate inversions of data at individual time step, and create difference models (e.g., percen-tage changes or difference ratios) between the results. This approach, while less computationally expensive,can omit or exaggerate important features that may be highlighted by a true time-lapse inversion.

The result of any inversion is typically an image showing the distribution of resistivity in the subsurface. Thefull process from acquisition to inversion is often referred to interchangeably as electrical resistivity tomogra-phy or electrical resistivity imaging. Tomographic inversion can be computationally expensive, and for thisreason, ER data are occasionally not inverted to produce images of the subsurface. Alternative uses of ER datainclude assessing the rawmeasurements, or apparent resistivity, visually and statistically to detect changes inthe subsurface over time. Another use of noninverted data is to look at transfer resistances between electro-des in an array. These measurements do not divulge information on spatial variations in the resistivity data,but are still sensitive to changes in moisture content over time. The advantage of these approaches is therapidity of assessments that can be made with minimal manual interpretation and computing resources.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 12

In the case study publications identified in Appendix A, 17 of the surveysused time-lapse ER methods; five were under controlled test conditions,four utilized transient measurements, and eight described semipermanentinstallations of electrode arrays. In most of the studies, the stated aim ofthe time-lapse ER surveys was to identify properties pertaining to watermovement or distribution within the landslide setting. These studies wereall undertaken in 2-D, with the exception of one time-lapse ER survey thatwas undertaken in 3-D (Uhlemann et al., 2017).

In one controlled test case study by Travelletti et al. (2012), simulated rain-

fall was applied to a 100-m2 area of the accumulation zone in the Lavallandslide, South French Alps. A 47-m-long profile comprising 48 electrodesspaced 1 m apart collected electrical resistivity data over a 67-hr period ofartificial rainfall of 11 mm/hr. The experiment started under unsaturatedconditions (at approximately 27% saturation) and aimed to determinethe time at which steady state flow was reached in the landslide body.By comparison with piezometer data, this was determined to haveoccurred 21 hr after the onset of artificial rainfall. The results of the time-lapse ER survey are shown in Figure 7. A total of 32 data acquisitions weremade during the experiment, resulting in an acquisition frequency ofapproximately 2 hr. The results of the time-lapse survey were able tomap the evolution of saturation within the slope, as well as determinethe bedrock geometry of the area under investigation. The authors recom-mended increased acquisition rates during the experiment to better moni-tor wetting processes, and the incorporation of 3-D arrays to improve thequantitative analysis of the data, primarily due to the suspected 3-D effectsthat are likely to have occurred during the experiment, but that are notcaptured by the 2-D inversion approach of the data.

The study by Travelletti et al. (2012) showed a relatively high frequency ofdata acquisition over a short period of time, and in this regard is similar toother artificial rainfall experiments which utilized time-lapse ER methods(Grandjean et al., 2009; Jomard et al., 2007; Lehmann et al., 2013). The timescales and acquisition rates of these studies are in contrast to longer-termstudies that utilize transient measurements and semipermanent installa-tions. In one study by Bièvre et al. (2012), transient time-lapse ER measure-ments were used to determine the role of fissuring on water infiltrationover a 498-day period, by acquiring four time-lapse ER data sets in theunstable clay slopes of the Avignonet landslide in the Trièves area of theFrench Alps, France. Although the campaign lasted over a year, resistivitydata sets were acquired at 0, 29, 236, and 498 days, showing an irregularacquisition schedule. Despite the relatively low-frequency and irregularityof acquisition, the time-lapse approach suggests that the observed evolu-tion of fissuring shows variations in resistivity that are related to the waterstorage capacity of the fissures. Higher-frequency data set acquisitionwould benefit the monitoring of the evolution of fissure-aided infiltration.

The use of semipermanent 3-D arrays and regular acquisition schedulesare able to overcome the issues of irregular acquisition rates and 3-Deffects in 2-D surveys, and pseudo-3-D surveys interpolated from 2-D datasets. These 3-D effects occur where a feature which may not be directlylocated beneath the ER profile influences the data. Such a study wasdescribed by Uhlemann et al. (2017), in which a semipermanent time-lapsemonitoring system was used to collect 3-D ER data approximately everytwo days at the Hollin Hill landslide in North Yorkshire, UK. The study

Figure 7. The results of a controlled test time-lapse ER survey by Travellettiet al. (2012) under artificial rainfall conditions (see text). The inverted ERimages showing a decrease in resistivity in response to the increased saturationof the subsurface. Modified from Travelletti et al. (2012). Copyright © 2011 byJohn Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 13

involved several other novel aspects beyond the incorporation of 3-D data acquisition. First, inversionmodelswere able to account for movement experienced by the electrodes in the sliding mass (Wilkinson et al., 2016),which can introduce significant errors in the data if not accounted for. Additionally, the resistivity data wereconverted to gravimetric moisture content (GMC) by fitting of a Waxman-Smits model (Waxman & Smits,1968) to laboratory experiments of wetting and drying of samples obtained from the site (Merritt et al.,2016). This allowed the production of volumetric images of GMC data, showing the hydrogeologicalvariation in the landslide body both before and after a slope failure in December 2012 (Figure 8).

Other long term time-lapse ER studies have utilized data in other novel ways in order to capture subsurfacemoisture variations within landslides over time. Studies utilizing noninverted ER data as an indicator of hydro-logical variation have beenmade by several authors. These studies were able to detect features such as watertable variations (Palis, Lebourg, Tric, et al., 2017), links between hydrological patterns and changes in appar-ent resistivity (Lebourg et al., 2010; Palis, Lebourg, Vidal, et al., 2017), and relationships between resistancedata and subsurface moisture dynamics (Merritt et al., 2018). In each of these studies, the data requiredsignificantly less processing than if tomographic inversion of the data had taken place.

After conducting time-lapse surveys to confirm the application of ER to the Vence landslide in south-easternFrance, Lebourg et al. (2010) used apparent resistivity correlations compared to rainfall and groundwaterlevels to conduct multidimensional statistical analysis and determine prerainfall and postrainfall states ofthe landslide over a 90-day monitoring period. The simultaneous presentation of individual apparent resistiv-ity measurements alongside rainfall events over a 275-day monitoring period at a landslide at Ampflwang-Hausruck, Austria, by Supper et al. (2014) allowed for rapid observations of the response of measurementsto rainfall events to be made without the need for time-lapse inversions. However, it is worth noting thatSupper et al. (2014) still produced time-lapse inversions for detailed determinations of movement eventsand relationships with precipitation. At the La Clapière landslide in the Alpes Maritimes, France, Palis,Lebourg, Vidal, et al. (2017) compared themedian apparent resistivity value from a semipermanent ER systemto a cumulative rainfall index, and observed both short-term decreases in resistivity associated with acuterainfall events, as well as longer seasonal variability linked to solar radiation. Apparent resistivity data froma 9.5-year monitoring period was statistically analyzed using cluster analysis in a study by Palis, Lebourg,Tric, et al. (2017). The results of the study allowed the determination of distinct hydrogeological units withinthe landslide without the need to make assumptions about the local geology, identifying areas of the land-slide that react differently to water infiltration. However, the authors ultimately recommend further work to

Figure 8. Volumetric images of gravimetric moisture content (GMC) derived from a 3-D time-lapse ER system (Uhlemann et al., 2017) showing the progressivedrainage of a landslide after significant movement in December 2012. Volumetric images shown here are from the later part of amonitoring campaign spanning overthree years, and show the changes in GMC relative to a baseline model acquired at the beginning of the monitoring period in March 2010. Modified from Uhlemannet al. (2017). Copyright © 2017 by John Wiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 14

establish petrophysical relationships between the resistivity and moisture content, similar to the approachesundertaken by Uhlemann et al. (2017). Merritt et al. (2018) utilized resistance measurements to conduct rapidobservations of moisture dynamics at the Hollin Hill landslide in North Yorkshire, UK. The resistance data werederived from the same ER system used in the study by Uhlemann et al. (2017). As resistance data can beanalyzed before the process of inversion, the study proved to be a rapid means of assessing the near surfaceof the Hollin Hill landslide to identify areas of potential preferential infiltration. This type of approach canprove useful for the determining areas of a landslide in whichmore computationally expensive investigationsshould be undertaken.4.4.2. Passive Geoelectrical Methods4.4.2.1. SPSelf-potential (SP) techniques measure the presence of naturally occurring charges in the Earth surface,primarily streaming potentials and electrochemical potentials, although thermokinetic potentials andcultural activity may also be recorded (Colangelo et al., 2006). As such, SP is defined as a passive geophysicalmethod. Potentials between pairs of electrodes are measured to create areal maps of variation inself-potentials. In the context of landslides, this is normally to determine areas of subsurface flow withinthe subsurface material. As with electrical resistivity, semipermanent arrays of electrodes can be installedto facilitate high-frequency, long-termmonitoring campaigns. Tomographic inversion of 2-D and 3-D surveyscan be undertaken to ascertain information in cross section and volumes (Ozaki et al., 2014).

Only one study used SP monitoring of landslides (Colangelo et al., 2006) in which temporal fluctuations of SPsignals were monitored over a 24-hr period during a period of precipitation. A linear array of electrodes weredeployed, and the data were inverted (to produce a tomographic inversion cross section) to show thedynamics of subsurface flow during the rainfall event. The study describes in a qualitative manner the varia-tion in flow, instead of moisture content as normally targeted by ER surveys. As the SP method does not giveany information in itself on geological characterization, the interpretations were made with the knowledgeacquired from previous SP mapping and ER investigations of the landslide setting.

4.5. Seismic Monitoring Case Studies

As with geoelectrical methods, seismic methods can be split between active and passive methods. Activeseismic monitoring records seismic signals generated by artificially produced sources. Passive methods tendto record seismic signals either generated by endogenous landslide activity (i.e., seismic signal produced bylandslide movement) or analyze ambient signals from exogenous sources.4.5.1. Active Seismic MethodsActive seismic methods provide information about the elastic properties of the media through which seismicwaves travel. This reveals geomechanical properties of subsurface materials and can indicate changes in thesubsurface stress conditions associated with movement, as well as providing information on the porosity ofmaterials, and in turn fluid flow characteristics.

Seismic refraction (SR) and surface wavemethods (SW) use field deployments of surface geophones linked toa seismograph to record properties of artificially generated seismic waves. In SR, the arrival times of compres-sional waves (P waves) and shear waves (S waves) are recorded. P and S waves are typically generated usingan impact source (e.g., a sledgehammer). P waves are generated by a vertical impact to a plate resting on theground surface, and recorded using geophones sensitive to movement in the vertical direction. S waves aregenerated by a horizontal impact to a plate coupled to the ground, and the resultant waves recorded by geo-phones sensitive to motion in the horizontal plane. The arrival times (“first breaks”) on each geophone giveinformation about the refracted pathway taken by the wave between the source and geophone. This infor-mation can then be inverted to give a model of the Pwave or Swave distribution in the subsurface, which canthen be interpreted in terms of variations in elastic properties of the subsurface.

SW methods record surface waves (primarily Rayleigh waves, although Love waves can also be analyzed),which are typically much slower and of greater amplitude than refracted P waves and S waves. Analyzingthe frequency content of these waves can be used to produce dispersion curves, which show phase velo-city as a function of frequency. By analyzing the fundamental mode energy in these images, a surface-wave velocity inversion is used to produce a 1-D model of S wave velocity as a function of depth.Lateral measurements of surface-wave velocity profiles can then be collated to produce 2-D sections ofinterpolated S wave velocity throughout the ground. Calculating S wave velocity from surface waves is

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 15

often preferred by investigators, as the field setup for acquiring surface waves is effectively the same asthat of a P wave SR survey, removing the need to repeat S wave SR surveys to obtain S wave velocities(Bergamo et al., 2016; Pasquet et al., 2015).

Two case studies utilized active seismic monitoring methods in landslide settings. Of these, one used SR sur-veys (Grandjean et al., 2009) and one utilized SWmethods (Bièvre et al., 2012). Both studies achieved the sta-ted aims of the experiments, but show relatively poor potential for obtaining long-term data sets at frequentintervals; indeed, no examples of semipermanent installations of active source SR or SW applied to landslidesare apparent in the literature. This is likely due to the relative complexity of remotely generating the neces-sary seismic sources, although analogous borehole systems do exist (Daley et al., 2007).4.5.1.1. SRIn the study of Grandjean et al. (2009), the main focus of the time-lapse P wave SR surveys was to determinefissure density during a simulated rainfall experiment on a landslide in the Laval catchment in Alpes-de-HauteProvence, France. The P wave survey was undertaken using a 48-channel seismic system, using 40-Hz geo-phones and a hammer source. The aim of the survey was to integrate the SR data sets with simultaneouslycollected ER data sets using a fuzzy logic approach. Direct analysis of the P wave SR suggested that low Pwave velocities in the near surface were due to increased fissure density. However, in contrast with the ERsurvey, P wave SR showed very little change with increasing saturation over the course of the experiment,which underscores the usefulness of integrated use of SR and ER surveys.4.5.1.2. Analysis of SWBièvre et al. (2012) used the variations in SW dispersion (Rayleigh waves) to assess the state of fissuring in theclay-rich slopes of the Avignonet landslide in the Trièves area of the French Alps, France. Surface waves wereacquired using 24 4.5-Hz geophones, and a hammer source. Analyses of the surface waves using amplitudesand spectral ratios were undertaken for two different acquisition dates, one in January and the other in July,in order to determine relative differences between the states of fissuring on each occasion. The authorsdetermined that the fissures remained open at these different times, with some possible minor variation inthe depth of the fissures.4.5.2. Passive Seismic MethodsIn order to obtain continuous seismic data over longer monitoring periods, it is necessary to look to methodsmore typically adopted by the seismological monitoring community for the assessment of large-scale naturalhazards such as volcanoes and earthquakes (Daskalakis et al., 2016). In recent years, the application of passiveseismic monitoring techniques have increasingly been applied to engineering issues (Öz, 2015), with notableadvances in the field of ambient noise measurements and the application of spectral analyses to passivemeasurements (Park & Miller, 2008; Planès et al., 2016). Early work by Suriñach et al. (2005) identified changesin the frequency content of seismic records at avalanche and landslide sites that were associated with initia-tion of mass movement.

Although the treatment and processing of passive seismic data vary depending on the intended use, theprincipal style of acquisition is essentially the same. A seismic recording instrument is left for a period of timeto record signals that may be generated by movement within the landslide body, or signals that are gener-ated by known or unknown, endogenous and exogenous sources. The instruments used can be geophones(typically for shorter recording periods), broadband seismometers, or other adapted sensors (e.g., acceler-ometers). Properly maintained and adequately powered semipermanent installations of seismic sensorsare able to record continuous data sets for very long time scales. For example, the monitoring period in astudy by Renalier, Jongmans, et al. (2010) exceeded 2.5 years. Some studies employ sensor deploymentsfor much shorter periods of time (i.e., <100 days; e.g., Amitrano et al., 2007), in what have been describedhere as short-term installations. These deployments are generally short due to equipment availability orthe lack of in-field power sources to maintain long monitoring periods. Single-sensor deployments can beused, for example, to detect movements within an area (e.g., Amitrano et al., 2007), or multiple-sensor deploy-ments can be used for more detailed interpretations (e.g., Brückl et al., 2013).

The case studies reveal a range of data analysis methods using passive seismic approaches. To simplify thedescription of approaches, the use of passive seismic methods is split into four categories: seismichorizontal-to-vertical ratio measurements (S-H/V), event detection, characterization and location (S-EDCL),seismic ambient noise cross-correlation (S-CC), and seismic ambient noise tomography (S-ANT).

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 16

4.5.2.1. S-H/VThe S-H/V technique involves recording ambient noise and analyzing the Fourier amplitude spectra ratiobetween the horizontal and vertical components of the seismometer, known as the horizontal-to-verticalratio (Chatelain, 2004; Nakamura, 1989). The resulting S-H/V spectral ratio curves can be used to determinerelative seismic impedance in the subsurface, and can therefore identify changes in layer thicknesses orthe presence of other features, such as slip planes.

S-H/V methods have been used in two studies to monitor landslides (see Appendix A), but with two very dif-ferent approaches. In Amitrano et al. (2007), S-H/Vmeasurements weremade from a short-term installation ofa single broadband seismometer located at the Super-Sauze landslide in France. The calculated cumulativeRMS of the S-H/V ratio for several frequency bands showed a correlation with periods of increased rainfall,indicating changes in the mechanical properties of sliding material during periods of acceleration.

Imposa et al. (2017) utilized a transient measurement approach, using a sensor specifically designed for theacquisition of S-H/V readings. These readings were acquired at the same stations in two separate surveysseparated by five years. For transient S-H/V measurements, recording durations between 2 and 24 min aretypically used (Acerra et al., 2002). The individual S-H/V spectral ratios that are produced per sensor locationwere processed and interpolated to produce cross sections of seismic impedance, showing changes in thelocation of the potential slip surface postfailure.

4.5.2.2. S-EDCLS-EDCL utilizes the more traditional aspects of seismology commonly applied in earthquake and volcanomonitoring. The approach is to detect events and classify them based on the source provenance and genera-tion. This approach involves direct interpretation of the signals recorded by semipermanent seismic sensors,either by identifying and classifying seismic events associated with movement in the landslide mass (i.e., “sli-dequakes” (Gomberg et al., 1995)) or by precursory activity that may be linked to impending movement (aphenomenon observed in rockfalls and other brittle landslides (e.g., Bell, 2018, Poli, 2017)), or a responseto recent increased infiltration (e.g., Palis, Lebourg, Vidal, et al., 2017). A single seismic sensor installation iscapable of achieving this type of monitoring (e.g., Amitrano et al., 2007); however, a network of seismometersis required to locate detected events (e.g., Walter et al., 2011). It is worth mentioning here that S-EDCL meth-ods are also the only approaches used in the near-real-time detection and characterization of landslides atthe regional scale (e.g., Kao et al., 2012; Manconi et al., 2016), and are commonly used in studies relating torockfalls and other fast-failing landslides.

An example of event detection and characterization is described in Provost et al. (2017). In this study, theauthors proposed an automatic classification system, in order to identify and separate seismic eventsrecorded at the Super-Sauze landslide in France based on their source and provenance. Sources included sig-nals from rockfalls and slidequake events within the landslide, and earthquakes and ambient noise eventsoriginating from outside the landslide setting. Example seismic signals are shown in Figure 9.

Brückl et al. (2013) undertook seismic monitoring of the Gradenbach landslide in Austria. They were able tolocate events recorded using six seismometers installed at the site, and noted increases in seismic activity upto 1.5 months before the manifestations of slope deformations which coincided with snowmelt. Palis,Lebourg, Vidal, et al. (2017) also noted increased seismic activity as a result of increased precipitation atthe La Clapière landslide, France, aiding in the establishment of rainfall threshold values for slope movementat the site.

The studies by Gomberg et al. (2011) at the Slumgullion landslide in the United States, Walter et al. (2011) atthe Heumoes landslide in Austria, and Walter et al. (2012) at the Super-Sauze landslide in France are all sum-marized in a study by Walter et al. (2013). The authors related increases in seismic activity to brittle failure (i.e.,fracture generation) of the landslide material, but also noted periods of movement that were not associatedwith increases in seismic activity. These aseismic movement events were related to periods of viscous creep,rather than brittle failure events. Location of the seismic sources of these events showed very different areasof seismic activity, and a strong association with the profile of the slip surface, the type of materials present,and the style of failure. Figure 10 shows the slidequake events located at the Heumoes landslide using S-EDCLmethods (Walter et al., 2013).

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 17

Figure 9. Example of seismic event detection and classification at the Super-Sauze landslide. The types of seismic events include endogenous (rockfall and slide-quake) and exogenous (earthquake and ambient noise) events. Modified from Provost et al. (2017). Copyright © 2017 by John Wiley & Sons, Inc. Reprinted by per-mission of John Wiley & Sons, Inc.

Figure 10. Locations andmagnitudes of slidequake (i.e., seismic activity generated by slopemovement) events detected atthe Heumoes landslide, Austria, by several seismic networks. Events are identified by different size and color dots on themap. The seismic networks include one semipermanent installation (see “permanent network”) and several short-termdeployments (see “field campaigns 2005–2008”). From Walter et al. (2013). Copyright © 2013 by Environmental andEngineering Geophysical Society. Reprinted by permission of Environmental and Engineering Geophysical Society.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 18

4.5.2.3. S-CCCross-correlating signals from a pair of seismic sensors recording thesame random ambient noise can be used to generate the impulsesignature (Green’s function) of wave propagation between sensors(Wapenaar et al., 2010). Many sources of natural and anthropogenicnoise can be utilized (for example, oceanic noise, vehicle traffic noise)if the source can be averaged over time to simulate a random ambi-ent source. These cross-correlation functions (CCFs) can be used tomonitor variations in surface-wave velocity over time, indicatingchanges in the stress field of subsurface materials (Lecocqet al., 2014).

Two of the case studies (Appendix A) utilize cross-correlation methods todetermine variations in geological material strength over time within mov-ing landslides. Renalier, Jongmans, et al. (2010) observed increases in seis-mic wave arrival times between two semipermanent seismometers over a32-monthmonitoring period at the Avignonet landslide in the Trièves areaof the French Alps, France. The differential in arrival times equated to analmost 0.2% decrease in velocity over time, with relative differences inground motion between the seismometers only able to account for anequivalent of a 0.06% change in seismic wave arrival times. The authorsconcluded that the increase in arrival times corresponded to an increasein “damage” (i.e., fissuring and destruction of soil fabrics) to the materialof the landslide. This area of damaged material was characterized by astand-alone Swave SR survey, in which the velocity of the damaged mate-rial was shown to be 2 to 3 times less than the surroundingundamaged area.

A study by Mainsant et al. (2012) over a shorter period of 146 days at thePont Bourquin landslide in Switzerland was able to detect variation in surface-wave velocities a few daysbefore significant slope failure on 19 August (Figure 11). A drop of 2% in surface-wave velocity occurred overa period of 20 days after heavy rainfall, with the surface-wave velocity dropping a further 5% in the sevendays preceding a significant failure at the landslide. The study highlights the applicability of long-term seis-mic monitoring to determining variations in subsurface material strength, and also the potential landslidepredictive capacity of seismic monitoring.4.5.2.4. S-ANTA frequency-time analysis of the outputs from cross correlations can allow for the generation of dispersioncurves (Bensen et al., 2007), which can in turn be inverted for the S wave velocity of the near surface (e.g.,Stork et al., 2018). Interpolation of the inverted dispersion curves between sensor pairs allows for the produc-tion of images similar to those seen in previous active seismic and geoelectrical techniques. Linear deploy-ments of sensors allow the acquisition of data in 2-D profiles, and studies have been able to imagelandslide bodies in 3-D using nonlinear arrays, although these have not incorporated a time-lapse elementyet (Renalier, Bievre, et al., 2010).

Only one case study produced these cross-section images by acquiring transient measurements on separatedates acquired from an array of 12 seismometers. In the study by Harba and Pilecki (2017), ambient noise datawere acquired for 60-min periods at two profiles using 12 seismometers. Measurements were made at theprofiles at three separate dates over seven months. The study took place at the Just-Tegborze landslide inNowy Sacz, southern Poland. The first acquisitions were made during a dry period in January 2015, the sec-ond acquisitions in March 2015 after the spring thaw, and finally during a very wet period in July 2015. The 1-D inversions of dispersion curves were performed on the cross-correlated signals between sensor pairs, andthe subsequent interpolated 2-D cross sections showed variations in the shear-wave velocity distributionthroughout the landslide mass between acquisition dates (Figure 12). The authors attribute the decreasingshear-wave velocity in the midsection of the landslide to the increasing moisture content of the landslideassociated with spring thaws and increased rainfall.

Figure 11. (a) The results from calculating changes in relative surface-wavevelocity over time via cross correlation of ambient noise seismic recordsby Mainsant et al. (2012). Decreases in velocity by 2% develop over 20 days(1), before a total decrease of 7% is observed (2) in the seven days precedinga significant failure (shaded gray area). (b) The cross-correlation coefficientand (c) the daily and cumulative rainfall. From Mainsant et al. (2012).Copyright © 2012 by John Wiley & Sons, Inc. Reprinted by permission of JohnWiley & Sons, Inc.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 19

4.6. Integrated Surveys

The preceding section has focused on the individual methods utilized for the geophysical monitoring of land-slides. One major limitation of geophysical investigations is the nonuniqueness of results obtained from sur-veys. The process of data inversion reduces this inherent ambiguity by providing spatial distributions ofgeophysical parameters. However, inversion of geophysical data is in itself a nonunique problem, as usuallyan infinite number of subsurface models can explain the data equally well and constraints need to be intro-duced to obtain the most likely subsurface distribution of the sought parameter. Although this nonunique-ness of both raw and inverted data can be addressed, one major limitation remains that an individual

Figure 12. The results of ambient noise tomography (ANT) undertaken on a series of transient measurements acquiredfrom seismometers by Harba and Pilecki (2017). Ground moisture is increasing from January to July, likely causing theobserved in decrease shear-wave velocity in the landslide body. From Harba and Pilecki (2017). Copyright © 2016 by JohnWiley & Sons, Inc. Reprinted by permission of John Wiley & Sons, Inc.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 20

method is normally only sensitive to one physical property in the subsurface, the distributions of which maybe variable in both time and space, and the occurrence of which may depend on several transient factors.4.6.1. Multiparametric Geophysical MonitoringMultiple methods can reduce the uncertainty in the assumptions made about the ground condition by com-paring the interpretations of each. Further quantitative analysis can be made by the use of joint inversionswhich incorporate physically constrained factors when creating models of the ground condition. Thisapproach is referred to here as “multiparametric geophysical monitoring.”

Relatively few examples are present in Appendix A, in which more than one geophysical method has beenutilized in a monitoring campaign. Multiparametric geophysical monitoring campaigns (i.e., utilizing morethan one acquisition method) have been used in three of the studies identified in Appendix A, with all usingER and SR methods (Bièvre et al., 2012; Grandjean et al., 2009; Palis, Lebourg, Vidal, et al., 2017).

In the study by Grandjean et al. (2009), ER and SR surveys were carried out on co-located survey profilesacross the Laval landslide in Draix, south French Alps. The results of the two independent surveys were fusedusing a fuzzy logical approach, utilizing probabilistic methods to test the validity of expert hydrological meta-hypotheses across the area of the monitored subsurface. This approach allowed for the independent resultsof each method to be combined during a controlled rainfall experiment, resulting in the quantification of thedevelopment of expected phenomena (e.g., presence of fissures, development of saturated zones, and areasof high porosity). This study is the most in-depth attempt to fuse data in such a way that the output of thestudy is more robust than the typical cross sections of property distributions normally found in geophysicalsurveys. The probabilistic approach of the study also pays significant attention to the reliability, or converselythe uncertainty, of the data, another area that can be overlooked in the application of geophysical methods.

In the studies by Bièvre et al. (2012) and Palis, Lebourg, Vidal, et al. (2017), the geophysical data sets are notinterpreted in such a joined quantitativemanner, but instead, all contribute to the overall qualitative interpre-tation of the landslide processes at play in the individual studies. For example, Bièvre et al. (2012) exploitedthe sensitivity of electrical resistivity methods to the presence of water-infiltrating fissures, and the effect ofattenuation on surface waves caused by the presence of fissures to determine extents of fissuring in the nearsurface of the clay-rich Avignonet landslide. The different properties assessed by the two techniques wereable to qualitatively confirm the hypothesis that fissuring played an important role in the infiltration of waterat the landslide, despite the independent acquisition and processing of the data sets. Similarly, Palis, Lebourg,Vidal, et al. (2017) were able to link changes in the annual climatic cycle to variations in seismic monitoringresponses and electrical resistivity values, allowing for a better understanding of the response of the landslideprocesses to external influences such as increased rainfall.4.6.2. Environmentally Coupled MonitoringIncorporation of environmental data (e.g., rainfall data, displacement rates) allows for the temporal variationsin geophysical monitoring data to be classified as being due to causative property changes in the landslide(e.g., increased saturation preceding failure) or resultant property changes (e.g., seismic responses generatedby movement of the landslide). Incorporation of these external data sources is referred to in this review as“environmentally coupled geophysical monitoring.” The importance of monitoring environmental fluctua-tions has been noted by several authors. Luongo et al. (2012) stated the importance of a weather monitoringstation at the location of a prototype semipermanent ER system, and similar local weather data werecollected in other studies (Uhlemann et al., 2017). Particularly with ER, external environmental variations,most notably thermal variations, have a strong influence on the data, and thermal effects need to becorrected (Lucas et al., 2017).

The displacement of landslides does, however, present a problem for ongoing monitoring using geophysicalmethods. As mentioned previously, the success of monitoring campaigns depends on being able to accu-rately spatially locate the datum at which measurements are being taken. In some types of geophysicalmonitoring approach, for example, when using single-seismometer monitoring for event detection, thedetailed location of the seismometer and its movement over time may not be of consequence to the aimof the monitoring (Amitrano et al., 2007). However, in some other methods, such as semipermanent installa-tions of ER arrays, movements in electrodes can have profound effects on the data quality, creating artifactsin the data if unaccounted for. This is typically overcome by periodic measurement of the movement of elec-trodes. These positional measurements need to be taken at a frequency at whichmovement can be captured.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 21

Recent advances in geophysical inversion have been able to recover electrode movements over time, andtherefore incorporate the movements in to the processing of ER data to prevent the introduction of data arti-facts by electrode movements (Boyle et al., 2017; Loke et al., 2017; Wilkinson et al., 2010; Wilkinson et al.,2015). This has been an important step in developing the autonomy of semipermanent installations, andhas greatly improved the data quality available from such campaigns (Uhlemann et al., 2017).

Establishing thresholds for landslide movement from geophysical data coupled with environmental data is acritical component of understanding subsurface process evolution. Palis, Lebourg, Vidal, et al. (2017) wereable to establish a rainfall threshold of 3.5 ± 1 mm/day above which the seismic sensors at the La Clapièrerecorded increased endogenous events associated with movement. Palis, Lebourg, Vidal, et al. (2017) alsoshowed that long-term analysis of rainfall with ER and S-EDCL methods allowed for the variable kinematicsof the landslide to be established, with the authors noting delayed responses in displacement rates asso-ciated with the intensities, durations, and repeat intervals of rainfall events. In many cases however, the deter-mination of thresholds from rainfall alone coupled to geophysical responses is unlikely to be successfulunless the infiltration rate is well understood, as the subsurface moisture dynamics that can vary greatlydepending on lithology, precipitation, temperature, etc., are not captured by such a simplified relationship.This is particularly apparent where effective rainfall (i.e., the amount of water that arrives at ground level,which is not inhibited by evapotranspiration) or effective infiltration (i.e., the amount of water reaching thesubsurface, which is not inhibited by evapotranspiration and surface runoff) is not taken into account. Afurther approach can therefore be to couple geophysical responses to monitoring measurements made fromthe subsurface, via the use of geotechnical monitoring methods.4.6.3. Geotechnically Coupled MonitoringEven with the incorporation of multiple geophysical methods to landslide monitoring, the sensitivity of thegeophysical method (or methods) is to the changes in the soil or rock property that is being monitored,rather than being a direct measurement of the property itself. For example, shear-wave velocity can beused to calculate shear strength of the subsurface when other data (i.e., density data) are incorporated,but it is not a direct measurement of shear strength. In order for geophysical measurements to be ofuse to end-users (who may not be geophysicists), the results need to be coupled to known parametersin the landslide environment. The relationship between geophysical measurements and the petrophysicalproperties of rocks and soils has most widely been applied in the hydrocarbon exploration industry(Barton, 2007). These coupling relationships are normally established by the careful selection and prepara-tion of representative samples of subsurface material, upon which laboratory tests to determine the rela-tionship between environmental influences (e.g., soil moisture), petrophysical properties (e.g., porosity,density), and a geophysical response (e.g., electrical resistance) can be established, as in the example ofMerritt et al. (2016). A study by Carrière et al. (2018) into the rheological properties of soil samples fromsix clay-rich landslides across Europe recommended the application of geophysical methods for determin-ing the properties of slide-to-flow dynamics. The main challenge to establishing reliable geophysical-petrophysical relationships is that of sample selection, and ensuring that samples collected and testedare representative of the conditions of subsurface materials, as landslides typically have highly heteroge-neous subsurface conditions (van Westen et al., 2006).

Upon the establishment of such relationships the ability to then compare the results of these geophysical-petrophysical relationships with data gathered from sensors installed in landslides is referred to here as“geotechnically coupled monitoring.” Lehmann et al. (2013) determined the relationship between soil moist-ure as measured by time domain reflectometer sensors at the same time as ER measurements during twosimulated rainfall tests. From this relationship they were able to produce cross sections of volumetric moist-ure content (VMC) and demonstrate the variations in VMC over time. In a study by Lucas et al. (2017), VMCwas measured directly by several sensors on site, and the data used to derive a relationship between ER mea-surements and VMC, allowing for reliable spatial subsurface mapping of VMC. The authors noted the benefitof using ER for determining VMC values at a large spatial resolution, but noted the need for care in calibratingthe ER measurements to determine VMC. In a similar study, Travelletti et al. (2012) were unable to establish arelationship between moisture content and ER data due to issues with nonuniqueness in the ER data, andscalability issues with using sparsely located moisture content measurements derived from point sensors.This issue of inherent heterogeneity and extrapolation of data from a sparse network of direct measurementsensor remains a challenge for geotechnically coupled geophysical monitoring.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 22

Uhlemann et al. (2017) established a relationship between ER values and laboratory determined values ofgravimetric moisture content (GMC) from samples collected at the landslide site rather than using measure-ments derived from sensors during the geophysical monitoring period, although retrospective comparisonbetween the ER-derived GMC and sensor-measured GMC was in good agreement. The authors were thenable to establish a failure threshold based on GMC, rather than rainfall, a measurement that better reflectsthe complex interactions between rainfall and subsurface moisture. In particular, estimated values of subsur-face moisture content have implications for soil suction, which can be one of the most important stabilizingforces in landslides (Toll et al., 2011).

The complexities of establishing such relationships in practice are identified by Hen-Jones et al. (2017), whoidentified that such geotechnically coupled relationships often display hysteretic behavior in relation to sea-sonal variations, adding a temporal heterogeneity to the already present spatial heterogeneity present inlandslide settings. An additional consideration is that, in a similar way that geophysical responses can becaused by a multitude of factors, geotechnical property variations can have a variety of causes, for example,the reduction of shear strength in soils due to loss of suction in wetting and drying cycles, and soil fabric dete-rioration across successive seasonal cycles (Hen-Jones et al., 2017).

5. Discussion5.1. Strengths in Geophysical Monitoring of Moisture-Induced Landslides

There are three areas in which the geophysical monitoring of landslides are particularly successful. These arethe following:

1. Providing high-resolution spatial information on landslide features (i.e., discrete spatial features).2. Providing high-resolution temporal information on the evolution of landslide processes (i.e., analyzing

how a subsurface property has changed over time and inferring the causal process).3. Calibrating and coupling data with other sources of environmental and geotechnical information.The first area, the provision of high-resolution spatial information on landslide setting features, is an exten-sion of the aforementioned benefits of geophysical methods in earlier sections (Jongmans & Garambois,2007; Mccann & Forster, 1990). The same advances in technology that have allowed for the establishmentof permanent monitoring installations have also allowed for greater measurement accuracy and resolution(Loke et al., 2013). The application of geophysical methods to image the subsurface between direct measure-ment locations is well established (Parsekian et al., 2015), and typically is where the use of near-surfacegeophysical methods lends most benefit to ground investigations. This can allow identification of featuresincluding the slip surface (Harba & Pilecki, 2017), analysis of the volumetric extent of disturbed materials(Uhlemann et al., 2017), and identification of deformation features and preferential infiltration pathways(Bièvre et al., 2012). Images, produced by tomographic inversion of data, can be assessed by comparativeimage analysis (e.g., Friedel et al., 2006), or through difference analysis compared to an initial image (e.g.,Lucas et al., 2017). Direct measurements from these sources is near essential in the accurate interpretationof geophysical images, highlighting the complementarity between well-sited direct measurements (the pla-cing of which can be optimized by geophysical methods) and geophysical images of the subsurface. The geo-physical images produced are further enhanced by the expansion from 2-D to 3-D data capture (Uhlemannet al., 2017) and relation to property distributions (Crawford & Bryson, 2018; Lehmann et al., 2013).

The second area of strength for monitoring relates to the progression of landslide processes. These processesinclude rapid moisture infiltration in response to acute rainfall (Travelletti et al., 2012), seasonal variations insubsurface moisture content (Uhlemann et al., 2017), short-term variations in ground saturation (Lehmannet al., 2013) and material strength (Mainsant et al., 2012) immediately preceding failure, measurements offlow in landslide bodies (Colangelo et al., 2006), and detection of landslide movement (Brückl et al., 2013).The temporal resolution of data that can be obtained from geophysical monitoring campaigns in many casescan match those of installed point sensors (i.e., daily or subdaily measurements), while in many cases stillproducing subsurface images that cannot be obtained from such sensor networks (Supper et al., 2014).

The third area is the one in which most development is required, but perhaps the area in which the moststands to be gained for monitoring of landslides. Demonstrated data-coupling approaches include linkingchanges in geophysical measurements to rainfall (Palis, Lebourg, Vidal, et al., 2017) and ground moisture

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 23

(Uhlemann et al., 2017) thresholds preceding failure, establishing property relationships between volumetricmoisture content and ER measurements (Lehmann et al., 2013; Lucas et al., 2017) and shear strength para-meters and ER measurements (Crawford & Bryson, 2018). It is in this area that geophysical monitoringapproaches can contribute most to slope-scale EWS, not only due to the spatial and temporal resolution ofdata but by linking these data to other areas of landslide study that can be understood by the wide rangeof stakeholders involved in the establishment, maintenance, and management of EWS.

S-EDCL approaches are well represented in the literature, but currently deal primarily with a retrospectiveanalysis of seismic signals generated by landslides failure. ER monitoring is particularly well suited to moni-toring precursory infiltration and subsurface movements and distributions of water, which is a key driver inthe destabilization of landslides through the reduction of matric suctions in partially saturated conditions(Toll et al., 2011). In addition to this, the application of ER monitoring has seen large interest not only inthe field of landslide investigation and monitoring (Perrone et al., 2014) but also in other areas of hydrogeo-physical research (Binley et al., 2015). This interest has driven technological advancements in the applicationof geoelectrical monitoring of landslides, something which is likely reflected in the increasing length ofmonitoring periods and frequency of data set acquisition over time. A geotechnically coupled approach isimportant in informing the predictive capacity of geoelectrical monitoring systems. When integrated withtransmission of data from site to stakeholders via telemetry, it can allow for near-real-time monitoring ofconditions as they move toward critical hydrogeological failure states.

In summary, the success of ER methods to monitoring of landslides is due to the longevity of the monitoringperiods that can be achieved (e.g., Palis, Lebourg, Tric, et al., 2017), the resolution at which data sets can beacquired (e.g., Supper et al., 2014), the use of semipermanent installations of geophysical monitoring infra-structure coupled with environmental and geotechnical sensor networks (e.g., Uhlemann et al., 2017), theability for the results to be related to subsurface properties pertinent to landslide processes (e.g., Crawford& Bryson, 2018; Lucas et al., 2017), and in some cases the ability to yield critical subsurface information withminimal postacquisition processing (e.g., Lebourg et al., 2010; Merritt et al., 2018).

5.2. Challenges to Geophysical Monitoring of Moisture-Induced Landslides

Of the methods assessed in this review, ER monitoring has the greatest potential for monitoring subsurfacematerial properties which may lead to slope failure. This is despite limitations surrounding power supplies toinstruments, the practical considerations of deploying an electrode array in topographically difficult terrain,and the limitation of high-resolution spatial information to the near surface. With regards to semipermanentER monitoring, these issues have mostly been addressed as many of the semipermanent ER systems in thisreview have been developed specifically for the purpose of landslide monitoring, while other approacheshave adapted existing equipment and methodologies. Further technological developments are still requiredin these approaches to optimize their application to landslide monitoring.

Active seismic methods (SR, SW) require the use of impulsive sources for generating seismic waves on site,limiting their use to transient measurements rather than semipermanent installations (e.g., Grandjeanet al., 2009; Renalier, Jongmans, et al., 2010) as no remote impulse generators exist for this application in sur-face deployed arrays (Daley et al., 2007). Other methods, such as SP, seem to be less used due to potentialambiguities in data interpretation, and the heavy reliance in determining subsurface conditions, such aslithology, from other sources. However, the single example of SP monitoring by Colangelo et al. (2006)demonstrates the value of SP approaches, particularly when included as part of a wider geophysical, environ-mental, and geotechnical landslide monitoring campaign (Perrone et al., 2004). The passive approachprovided by SP monitoring should lend itself to integration into new and existing geophysical monitoringobservatories, as the infrastructure required is significantly less than that required for the establishment ofER monitoring systems.

While passive seismicmethods arewidely used to retrospectively analyze the patterns of seismicity associatedwith landslide movements, their use to monitor the evolution of subsurface conditions preceding failure isunderutilized. This can be considered amissed opportunity, particularly given their ability tomonitor at highertemporal resolutions than active geophysical methods, and image to greater depths (where such imagingcapabilities are able to be employed, such as using S-ANT). This is despite the field of passive seismic monitor-ing being both highly developed and widely applied in monitoring both macroseismic (e.g., Öz, 2015)

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 24

and microseismic (e.g., Chambers et al., 2010) events in a wide range of applications, including hydraulicfracture stimulation, geothermal systems, waste storage (e.g., CO2), and mining. Seismic monitoring is alsonot uncommon in landslide monitoring, most commonly for detection and analysis of rockfall failure events(e.g., Hibert et al., 2017; Manconi et al., 2016; Partsinevelos et al., 2016). Similar approaches have been under-taken in some of the case studies identified in this review, such as event classification (Palis, Lebourg, Vidal,et al., 2017) and event location (Brückl et al., 2013).

A significant challenge common to most geophysical monitoring applications remains the large volumes ofdata that can be acquired and the time requirements to quality assess, process, and interpret these data.Currently, most data are still assessed and processed manually, although significant strides in acquisitionmethods, automated filtering, and inversion methods have increased productivity in these areas (Lokeet al., 2013). Challenges such as the accounting for movements in electrodes in active landslides(Wilkinson et al., 2010; Wilkinson et al., 2015) have been overcome, although similar challenges remain inthe areas of automated image and pattern analysis (Chambers et al., 2015; Ward et al., 2016), the develop-ment of which would significantly reduce the requirement for manual processing of monitoring data.While robust hardware for geophysical monitoring seems to have reached good operational standards inrecent years, developments in the associated software capabilities are still required to optimize themonitoring process.

5.3. Future Look and Opportunities

Currently, the only other methodology that is able to deliver robust monitoring capabilities and soil and rockproperty identification in a similar manner to semipermanent ER monitoring systems is long-term passiveseismic monitoring campaigns. A particularly novel but promising approach to passive monitoring is theuse of ambient noise monitoring at landslide sites, with studies by Harba and Pilecki (2017), Mainsant et al.(2012) and Renalier, Jongmans, et al. (2010) demonstrating particularly interesting results. The predictivecapacity of using cross correlations of ambient noise is made clear in the study by Mainsant et al. (2012),where relative velocities in surface waves are seen decreasing days before slope failure. Critically, this failuredid not seem to be linked to obvious changes in the rainfall regime, suggesting that a long-term retrospectiveestablishment of threshold values from rainfall data would not have worked in this landslide setting. The pro-duction of subsurface images of S wave velocity from ambient noise recordings by Harba and Pilecki (2017)also shows that passive monitoring methods can determine subsurface properties and produce images atsimilar spatial resolutions to transient SR and SW surveys. However, improvements to the methodologydescribed in Harba and Pilecki (2017) could improve the robustness of the results. Ambient noise sources ide-ally need to be both stable in time and nondirectional, although the former can be overcome by averagingover increased measurement times (e.g., Mainsant et al., 2012). In the case study by Harba and Pilecki (2017),the acquisition of 60-min data sets derived from traffic noise may not be sufficient to reliably reconstruct theGreen’s function of an ambient noise field. Accurate reconstruction of the Green’s function is crucial due tothe accuracy of the imaging capability. The deployment of permanent arrays, from which longer ambientnoise records could be extracted, would greatly improve the reliability of the results. Despite this, the casestudy by Harba and Pilecki (2017) is one of the only attempts to create subsurface images from ambient noisefor the purpose of monitoring landslides. As such, it represents an important step in providing increased tem-poral and spatial subsurface geophysical models of landslides, and demonstrates the capacity for employingsuch approaches in future studies.

Crucially, unlike transient SR and SW surveys, subsurface images produced from ambient noise have thepotential to be produced at a similar acquisition frequency to long-term semipermanent ER monitoring sys-tems. Recent developments in technology, particularly around linear Distributed Acoustic Sensing monitor-ing, show the potential for these new senor types to replicate the resolution of traditional transient SWsurveys, but at much higher data acquisition frequencies (Dou et al., 2017). Developments in this field, poten-tially into 3-D surveys, will be particularly relevant to the monitoring of landslides. Renalier, Bievre, et al.(2010) demonstrate the application of these ambient noise applications using a network of seismometersto create a 3-D image of a landslide body, but without a time series element to the data. Similarly, while somenonmonitoring studies have analyzed seismic data and inferred property distributions from active seismicmeasurements (Uhlemann, Hagedorn, et al., 2016), this has not been applied to time-lapse seismic studies,actively or passively acquired, and represents a significant gap in the field.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 25

Therefore, there appears to be three areas in which seismic monitoring canbe improved for the application of monitoring landslides: by utilizingambient data sets acquired by dense seismometer networks orDistributed Acoustic Sensing arrays over longer periods of time, byattempting to link changes observed in these data sets to subsurface prop-erties, and by producing high-resolution spatial images of the subsurfacefrom passive data. In the past, a barrier to improving resolution of ambientnoise surveys has been the cost associated with broadband seismometerstypically used for very sensitive planetary-scale measurements. However,recent studies in other areas show the successful deployment of low-cost,low-energy consumption geophone-based networks for the monitoring ofglaciers (Martinez et al., 2017), infrastructure (Olivier et al., 2017;Salvermoser et al., 2015), and landslides (Deekshit et al., 2016). Similarmoves toward lower-power, lower-cost systems in ER monitoring are alsoapparent (Chambers et al., 2016), which will increase the availability andapplication of geophysical monitoring in the field.

The simultaneous acquisition of continuous passive seismic and ER datawould open up opportunities for the further development of integratedquantitative processing and inversion, further increasing the usefulness

of geophysical monitoring data. The near-real-time element of some monitoring systems could thereforebe used for near-time modeling and decision making on high-risk slopes endangering critical social or eco-nomic infrastructure, or risk to human life.

5.4. The Contribution of Geophysical Methods to Landslide EWS

The compilation of catalogs of landslide events is often the first step to constructing landslide inventorymaps. These maps inform the designation of zones based on their level of risk to landslide events (Fellet al., 2008; van Westen et al., 2006). Implementation of appropriate land use planning and subsequent enfor-cement of this planning is arguably the most effective tool for mitigating future risk from landslide events,and can be practiced at different scales (Cascini, 2008). However, in some situations, the presence of risk fromlandslides may exist in the absence of appropriate risk zonation maps, or may become apparent in the laterdevelopment of land. Changing environmental conditions can also induce new landslide events, or reactivatedormant landslide activity (Gariano & Guzzetti, 2016). In these instances, different courses of action must betaken to mitigate the risk from future landslide events.

In cases where the elements at risk of a landslide hazard cannot be relocated, or are of high socioeconomicvalue (Manconi & Giordan, 2015) or where structural mitigation measures would be too resource intensive,undesirable, or ineffective, the implementation of early warning systems (EWS) may be the preferred optionfor risk mitigation (Intrieri et al., 2012). EWS applied to natural hazards can be defined as “… monitoringdevices designed to avoid or to mitigate the impact posed by a threat” (Medina-Cetina & Nadim, 2008).Slope-scale EWS for landslide monitoring requires a detailed understanding of the triggers and mechanismsin order to be effective (e.g., Manconi & Giordan, 2016). Successful slope-scale EWS must include a detailedunderstanding of the character of the landslide, the processes acting on the slope, and the response of theslope to external influences, such as increased infiltration (SafeLand, 2012). The main components of EWSare shown in Figure 13.

Typically, slope-scale EWS are established by acquiring and monitoring changes in both environmental (i.e.,rainfall, temperature, soil moisture content, pore water pressure) and kinematic (i.e., slope movement) data(Uhlemann, Smith, et al., 2016). Commonly, these data that are acquired through geomorphological map-ping, geodetic surveys, and borehole measurements (e.g., inclinometers) are used in monitoring campaigns.Data can also be acquired from sensors measuring piezometric head, tilt, and soil suction. More recently,wireless sensor networks comprising sensors with low-power requirements and large data storage or tele-metric capabilities have been deployed in landslide settings to acquire direct measurement data with higherspatial and temporal resolution (Ramesh, 2014). Despite these advances, logistical and economic constraintsremain barriers to optimum data acquisition for the establishment of effective EWS. Currently, geophysicalinvestigations and monitoring methods are not widely used in slope-scale landslide EWS.

Figure 13. Four main activities of landslide early warning systems (EWS)reproduced from Intrieri et al. (2013). Sections underlined show the areasin which geophysical investigations and/or geophysical monitoring can beused to assist in landslide EWS.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 26

Table 3 shows the contribution that geophysical monitoring approaches can make to the four stages ofestablishing landslide EWS, as defined by Intrieri et al. (2013) and shown in Figure 13. Entries highlightedgreen are those areas in which capability has been demonstrably developed. Orange entries indicate thoseareas in which challenges still exist but where the requisite technology may have been applied to other areasof geophysical monitoring outside of landslide studies, for example, in the automatic detection of geophysi-cal responses, and installation of high-resolution monitoring arrays. Entries in red indicate areas in which sig-nificant development is still required before operational deployment to a field setting, the most significantone of which being the ability to issue reliable warnings based on the integration of environmental, geotech-nical, and geophysical data.

6. Conclusions

The use of geophysical monitoring systems to assess the evolution of subsurface property distributions andchanges over time have been reviewed. The benefit of long-term geophysical monitoring campaigns inassessing the evolution of MIL is demonstrable. A significant advantage of the use of geophysical monitoringsystems is their complementarity with existing monitoring systems. Many areas of landslide monitoring arehighly developed, with innovative advances in technology progressing a wide range of applications.Developments in UAV technology (Walter et al., 2009), interferometric synthetic-aperture radar (Hilleyet al., 2004), and other remote sensing capabilities (Kirschbaum et al., 2015) have increased the temporaland spatial resolution at which surface deformations can be monitored. Geophysical data can enhance thedeformation data captured by these technologies by providing an understanding of the subsurface pro-cesses driving them, something that is not achievable at such high temporal and spatial resolutions withother monitoring methods. Sensor technology has also advanced, with installations of tiltmeters, inclin-ometers, and piezometers being enhanced with the addition of shape acceleration arrays, acoustic-emissionwaveguide sensors, and tensiometers (Uhlemann, Smith, et al., 2016). Geophysical methods are able to inter-polate between the locations of these direct measurements, extending by proxy the spatial coverage of thesesensor types.

Key advances in the field of geophysical monitoring in the last decade include the following:

Table 3The Contributions That Geophysical Monitoring Approaches Can Make to EWS in the Context Proposed by Intrieri et al. (2013) in Figure 13

EWS Activity(After Intrieri et al., 2013) Contribution by Geoelectrical Monitoring Methods Contribution by Passive Seismic Monitoring Methods

Monitoring Instrumentsinstallation

Installation of semipermanent 2-D monitoring arrays Installation of seismometer(s)

Installation of semipermanent 3-D monitoring arrays Installation of dense networks of low-cost, low-powerseismic sensors or DAS arrays

Data collection Regular, short-time scale (≤1 day) data acquisition Continual data recording

Large data storage or telemetric capabilities

Data transmission Telemetric capability can send data to central observatory

Data elaboration Integration of data with environmental, geotechnical and kinematic data collected simultaneously from locally installedsensors

Forecasting Datainterpretation

Automatic image analysis and change detection of hydrogeologicalprocess through time-lapse imaging

Automatic seismic event detection

Automated geotechnical coupled and environmentally coupledmonitoring

Automated event classification

Automated event location

Comparison withthresholds

Establishment of relevant subsurface hydrogeological thresholds precedingmovement (e.g., moisture content, shear strength)

Forecast methods Monitoring of subsurface conditions toward hydrogeological failurethreshold

Monitoring of subsurface conditions towardgeomechanical failure threshold

Warning High-resolution spatial and temporal geophysical monitoring, calibrated with and linked to environmental, geotechnical, andkinematic data to provide multilevel warnings based on monitoring of several physical subsurface parameters

Note. Those contributions highlighted red indicate areas in which the most development is still required, those in orange are where research is currently beingundertaken but not yet applied to the monitoring of MIL, and green identifies those areas in which geophysical monitoring can already contribute the slope-scaleEWS on MIL.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 27

Table A1Published Case Studies Utilizing Geophysical Monitoring of Landslides Since 2006 With Information on the Landslide Setting

AuthorsLandslide nameand location

Approximatedepth to slide

(m bgl)Approximateslope angle (°)

Approximatearea (m2)

Movement type(after Hungr et al.,

2014) Material

Brückl and Mertl(2006)

Gradenbach 1.68x106 Mountain slopedeformation

Phyllites and calcareous micaschistsSchober Range,

Carinthia, Eastern Alps,Austria

Brückl and Mertl(2006)

Hochmais-AtemskopfÖtztaler Alpen, Tyrol,

2.82x106 Mountain slopedeformation

Paragneisses, micaschists,orthogneisses, amphibolites

Eastern Alps, AustriaBrückl and Mertl(2006)

Niedergallmigg-Matekopf

2.64x106 Mountain slopedeformation

Pelitic genisses, micaschists

Samnaun Range, Tyrol,Eastern Alps, Austria

Colangelo et al.(2006)

Varco d’Izzo 15 – 25 8 – 16 1.82x105 –5.88x105

Compositerotational-translational -flow

Clay-marl terrains, includingblocks of marls, calcarenitesand limestones

Potenza, Basilicata,Southern Apennine,Italy

Friedel et al.(2006)

Toessegg 0.6 – 1.5 20 – 30 - Translational Quaternary deposits overlyingsandstoneBanks of Rhine,

Rüdlingen, SwitzerlandAmitrano et al.(2007)

Super-Sauze 5 – 10 25 - Slide - flow Weathered marl (clay)Barcelonette Basin,Southeast Alps, France

Jomard et al.(2007)

La Clapière 10 40 - Translational -rotational

Fractured gneissAlpes Maritimes, France

Bell et al. (2008) Lichenstein-Unterhausen

ILEWS Project update.

Swabian Alps,southGermany

Walter andJoswig (2008)

Heumoes - 9x105 Slide Loamy scree and till (silty,clayey and sandy material)Vorralberg Alps, Austria

Grandjean et al.(2009)

Laval 5 30 4x103 Slide-flow? Weathered marlsLaval catchment, ORE,Draix, South French Alps

Walter et al.(2009)

Super-Sauze Described in more detail by Walter et al. (2012)Barcelonette Basin,Southeast Alps, France

Helmstetter andGarambois (2010)

Séchilienne - 35 - 40 - Rockslide Mica schists with interbeddedquartz-feldspar layersBelledonne massif,

French Alps, FranceLebourg et al.(2010)

Vence 12 12 – 14 8.75x104 Translational Sandy-clay overlying marlsAlps Maritimes, SouthEast France

Renalier et al.(2010a)

Avignonet 5 – 42 8 – 15 1.5x106 Translational -rotational

Lacustrine and alluvial clayTrièves area, FrenchAlps, France

Gomberg et al.(2011)

Slumgullion 20 - 1.17x106 Translational Sandy, silty clay and areas ofbouldery debris, clay andpond and stream sediment

San Juan mountains,Colorado, United States

Lacroix andHelmstetter (2011)

Séchilienne - 35 - 40 - Rockslide Mica schists with interbeddedquartz-feldspar layersBelledonne massif,

French Alps, FranceWalter et al. (2011) Heumoes - - 9x105 Slide Loamy scree and till (silty,

clayey and sandy material)Vorralberg Alps, Austria

Bièvre et al. (2012) Avignonet 5 – 42 8 – 15 1.5x106 Translational -rotational

Lacustrine and alluvial clayTrièves area, FrenchAlps, France

Luongo et al. (2012) - - - - Translational -rotational

Schist, flyschPicerino region,Basilicata region, Italy

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 28

Table A1 (continued)

AuthorsLandslide nameand location

Approximatedepth to slide

(m bgl)Approximateslope angle (°)

Approximatearea (m2)

Movement type(after Hungr et al.,

2014) Material

Mainsant et al.(2012)

Pont Bourquin “few metres”to 11

- 8x103

(potentially 3x104

– 4x104

Active compositeearthslide - earthflow

Triassic dolomites andgypsum, with flysch depositsmid-slope, including siltstonesand conglomerates

Switzerland

Travelletti et al.(2012)

Laval 1 – 6 32 4x103 Slide-flow? Weathered marlsLaval catchment, ORE,Draix, South French Alps

Walter et al. (2012) Super-Sauze 5 – 10 25 - Slide - flow Weathered marl (clay)Barcelonette Basin,Southeast Alps, France

Brückl et al. (2013) Gradenbach - - 1.7x106 Phyllite, schists andcarbonatesSchober Range,

Carinthia, Eastern Alps,Austria

Lehmann et al.(2013)

Controlled site 0.7 – 5.6 38 262.5 Slide? Silty sandy clay overlyingmarlstones and sandstonesBanks of Rhine,

Rüdlingen, SwitzerlandTonnellier et al.(2013)

Super-Sauze 5 – 10 25 - Slide - flow Weathered marl (clay)Barcelonette Basin,Southeast Alps, France

Tonnellier et al.(2013)

Valoria 15 – 30? - 1.6x106 Translational slide? Flysch and clay-shaleNorthern Appenines,Dolo River Basin, Italy

Walter et al. (2013) Heumoes - 9x105 Slide Loamy scree and till (silty,clayey and sandy material)Vorralberg Alps, Austria

Walter et al. (2013) Slumgullion Study described in Gomberg et al. (2011)San Juan mountains,Colorado, United States

Walter et al. (2013) Super-Sauze Study described in Walter et al. (2009)Barcelonette Basin,Southeast Alps, France

Supper et al. (2014) Ampflwang-Hausruck 20 – 30 - 4.4x103 Rotational-translational

Quaternary colluvium,anthropogenic deposits andNeogene

Ampflwang, HausruckHills, Austria

Supper et al. (2014) Bagnaschino 8 - 1.5 x 105 Rotational-translational

Colluvium overlyingamphibolite and schistTorre Mondovì, Casotto

Valley, Cuneo/ PiedmontProvince, Italy

Kremers et al.(2015)

Badong YANGTZE-GEO project update.China

Jongmans et al.(2015)

Pont Bourquin Summary of Mainsant et al., 2012.Switzerland

Gance et al. (2016) Super-Sauze ~5? 25 - Slide - flow Weathered marl (clay)Barcelonette Basin,Southeast Alps, France

“a fewmetres”

Harba and Pilecki(2017)

Just-Tegoborze 10 – 12 and14 - 17

- - - Colluvial deposits overlyingflysch depositsNowy Sacz, Southern

PolandXu et al. (2016) Bank of Zagunao River 30 – 70 25 -35 1.06x106 - Colluvial deposits overlying

Devonian phylliteLixian County, SichuanProvince, China

Imposa et al. (2017) Tripi, north-easternSiciliy, Italy

4-Mar - 2.49x103 Translational? Colluvial deposits overlyingparagneiss

Lucas et al. (2017) Canton of Valais, SwissAlps, Switzerland

3-Jan 33 - 43 - - Gravels, silt and sand overlyingquartzite bedrock

Palis et al. (2017a) La Clapière <100-200? - >8x105 Rockslide Granodiorite and migmatiticgneissAlpes Maritimes, France (Jomard et al.,

2010)Palis et al. (2017b) Vence 15-Oct 14-Dec 8.75x104 Translational Sandy clay overlying Jurassic

limestone(Lebourg et al., 2010)

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 29

1. Telemetric capabilities in geoelectrical and seismic monitoringmethods, increasing the length of monitor-ing periods and allowing near-real-time data acquisition.

2. Coupling of geophysical data to geotechnical properties, particularly in the field of ER.3. The potential for integration of passive seismic and ER methods to decrease uncertainty in single-method

geophysical monitoring campaigns.A key area for future development is the implementation of multiparametric monitoring with geotechnicallycoupled approaches. This can be achieved through better calibration of geophysical measurements withgeotechnical data, and the establishment of threshold values for failure based on these outputs.

Recent advances in incorporating slope displacements in the data processing phase of geophysical monitor-ing campaigns has greatly reduced the presence of artifacts in the resultant subsurface images (Wilkinsonet al., 2010; Wilkinson et al., 2015), as well as advances in the inversion of time-lapse data (Loke et al.,2013) and automated image analysis (Chambers et al., 2015; Ward et al., 2016). The establishment of site-specific geotechnical-geophysical relationships to obtain the subsurface spatial distribution of geotechnicalproperties, and the ability to monitor changes in these properties over time, is valuable to a wide range ofend-users and stakeholders (Crawford & Bryson, 2018; Merritt et al., 2016; Uhlemann et al., 2017). Thedevelopments highlighted in this review underscore the potential for the combined use of geophysicalmonitoring methods with existing remote sensing, Earth observation, and direct measurement monitoringapproaches to improve the current capacity for high-resolution slope-scale landslide EWS (Intrieri et al.,2012; Intrieri et al., 2013).

Appendix A

This table lists all the case studies identified as part of this review. Studies were selected for their use of geo-physical monitoring approaches to study landslides as (see text). Where possible, further information regard-ing the landslide setting has been extracted from the case study text. A brief summary of the key findings ofthe study is found in the “notes” column (Table A1).

Glossary

Cascading Disaster: As defined by Pescaroli and Alexander (2015), these are “… extreme events, in whichcascading effects increase in progression over time and generate unexpected secondary events of strong

Table A1 (continued)

AuthorsLandslide nameand location

Approximatedepth to slide

(m bgl)Approximateslope angle (°)

Approximatearea (m2)

Movement type(after Hungr et al.,

2014) Material

Alps Maritimes, SouthEast France

Provost et al. (2017) Super-Sauze ~5? 25 - Slide - flow Weathered marl (clay)Barcelonette Basin,Southeast Alps, France

“a fewmetres”

Uhlemann et al.(2017)

Hollin Hill 3-Feb 12 3.9x104 Composite earth-slide earth-flow

Weathered mudstones andsandstonesNorth Yorkshire, UK

Crawford andBryson, 2018

Doe Run <8 - 2.5x103 Translational Silty clay colluvium overlyingshale and limestoneKentucky, USA

Crawford andBryson, 2018

Herron Hill <3 - 3.75x104 Composite rotationaltranslational?

Silty Clay colluvium andweathered shaleKentucky, USA

Merritt et al., 2018 Hollin Hill 3-Feb 12 3.9x104 Composite earth-slide earth-flow

Weathered mudstones andsandstonesNorth Yorkshire, UK

Note. ER electrical resistivity; SW surface wavemethods; SP self-potential; SR seismic refraction; S-EDCL continuous seismicmonitoring, including use of event char-acterization, detection, and location methods; S-H/V seismic monitoring utilizing horizontal-to-vertical ratio methods; S-CC seismic monitoring utilizing cross-cor-relation methods; S-ANT seismic monitoring utilizing ambient noise tomography methods.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 30

Table A2Published Case Studies Utilizing Geophysical Monitoring of Landslides Since 2006 With Information on Geophysical Monitoring

Authors

Geophysicalmonitoringmethod

Monitoringtype Dimension

Monitoringperiod(Days)

Number ofdatasets/

seismometers* Survey design Notes

Brückl andMertl (2006)

S-EDCL Semi-permanent

15 8* Detected, characterised andlocated seismic events on theslope-scale at a slowly deformingrock mass.

S-EDCL Semi-permanent

1 9*

S-EDCL Semi-permanent

10 8*

Brückl andMertl (2006)

S-EDCL Semi-permanent

? 5*

S-EDCL Semi-permanent

19 2*

S-EDCL Semi-permanent

5 11*

Brückl andMertl (2006)

S-EDCL Semi-permanent

9 5*

Colangelo et al.(2006)

SP Controlledtest

2D 1 8 1 x 50m profile (11electrodes at 5m spacing

Only SP monitoring case study.Showed time-lapse SPtomograpic images indicatingsubsurface flow. The 24 hourmonitoring period was part of alonger semi-permanentmonitoringcampaign.

Friedel et al.(2006)

ER Transient 2D 336 2 1 x 24.5m profile (50electrodes at 0.5m spcaing

Comparative analysis of ERtomographic images acquiredone year apart showedinfluence of rainfall.

Amitrano et al.(2007)

S-H/V Semi-permanent

1D 13 1* 1 seismometer H/V ratio from a singleseismometer shown in responseto landslide movement.

Jomard et al.(2007)

ER Controlledtest

2D 2 14 1 x 141m profile (48electrodes at 3m spacing

Time-lapse ER tomographicimages showed moisture changesover period of controlled rainfallexperiment.

Bell et al. (2008) ILEWS Projectupdate.

Walter andJoswig (2008)

S-EDCL Semi-permanent

1D 7 8 Two tripartite sensorarrays deployed in eachperiod, comprising of 1 x3-component sensorsurrounded by 3 x 1-component sensors

Two monitoring periods in rapidsuccession (nine daysseparation) showed increase innanoseismic activity five to26 hours after rainfall event.

S-EDCL Semi-permanent

7 8

Grandjean et al.(2009)

ER Controlledtest

2D 2.8 32 1 x 47m profile (48electrodes at 1m spacing

Controlled rainfall experiment,presented statistical time-laspeimages produced from ER andSR data. Same experiment as in(Travelletti et al., 2012).

SR Controlledtest

2D 2.8 23 1 x 47m profile (48geophones at 1m spacing

Walter et al.(2009)

Described inmore detail byWalter et al.(2012)

Helmstetterand Garambois(2010)

S-EDCL Semi-permanent

1D 715 45 3 x sensor arrayscomprising of 6 verticaland 1 3 component arrays

Two years of seismic eventdetection, classification anddetection showed a weak, but

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 31

Table A2 (continued)

Authors

Geophysicalmonitoringmethod

Monitoringtype Dimension

Monitoringperiod(Days)

Number ofdatasets/

seismometers* Survey design Notes

(all replaced bybroadband sensors in2009) and an additional 24channel geophone arrayinstalled in 2008.

significant, correlation of rockfallevents with rainfall, with even1mm of rain showing an increasein seismically-detected rockfallevents, although events stilloccurred in the absence ofrainfall.

Lebourg et al.(2010)

ER Semi-permanent

2D 90 90 1 x 115m profile (24electrodes at 5m spacing

Used statistical analysis ofresistivity data to correlate withrainfall.

Renalier et al.(2010a)

S-CC Semi-permanent

2D 944 2* 2 seismometers fromlarger array

Cross-correlation betweenseismometer pair demonstratedlandslide movement by usingseismic arrival times.

Gomberg et al.(2011)

S-EDCL Semi-permanent

2D 9 88* Some issues with instrumentbatteries in achieving fullmonitoring coverage withcomplete network, but at least 2days complete monitoringachieved. Determined movementoccurs both seismically andaseismocally at the Slumgullionlandslide.

Lacroix andHelmstetter(2011)

S-EDCL Semi-permanent

2D 425 45 3 x sensor arrayscomprising of 6 verticaland 1 3 component arrays(all replaced bybroadband sensors in2009) and an additional24 channel geophonearray installed in 2008.

Focuses on locating seismicevents within the Séchiliennerockslide associated with rockfallsand basal movements, the lattercausing microearthquakes. Someareas showed aseismicmovements.

Walter et al.(2011)

S-EDCL Semi-permanent

<28 8* Two tripartite sensorarrays deployed,comprising of1 x 3-component sensorsurrounded by3 x 1-component sensors

Extension of Walter and Joswig(2008). Monitoring periods of oneto four weeks were undertaken in2005, 2006, 2007 and 2008.Showed continued increase ofseimsic activity five to 26 hoursafter rainfall events, presumed tobe linked to moisture-inducedfailure.

S-EDCL Semi-permanent

<28 8* Two tripartite sensor arraysdeployed, comprising of1 x 3-component sensorsurrounded by 3 x 1-component sensors

S-EDCL Semi-permanent

<28 12* Three tripartite sensorarrays deployed,comprising of1 x 3-component sensorsurrounded by3 x 1-component sensors

S-EDCL Semi-permanent

<28 20* Five tripartite sensorarrays deployed,comprising of1 x 3-component sensorsurrounded by3 x 1-component sensors

Bièvre et al.(2012)

ER Transient 2D 498 4 1 x 31.5m profile (64electrodes at 0.5m spacing

Focused on hydrologicalinfiltration through fissures.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 32

Table A2 (continued)

Authors

Geophysicalmonitoringmethod

Monitoringtype Dimension

Monitoringperiod(Days)

Number ofdatasets/

seismometers* Survey design Notes

Fissures identified through time-lapse ER tomographic images,and size of fissures over timeassessed using SW methods.

SW Transient 1D 198 2 1 x 57.5m profile (24geophones at 2.5mspacing

Luongo et al.(2012)

ER Semi-permanent

2D 146 584 1 x 47m profile (48electrodes at 1m spacing

Preliminary study of time-lapseER monitoring instrument onshallow landslide.

Mainsant et al.(2012)

S-CC Semi-permanent

2D 146 2* 2 seismometers Used cross-correlation of ambientnoise records to detect relativedecreases in landslide materialstrength before failure.

Travelletti et al.(2012)

ER Controlledtest

2D 2.8 32 1 x 47m profile (48electrodes at 1m spacing

Identified steady-state flowconditions in a controlled rainfallexperiment using time-lapse ERtomographic images, andestimated steady-state flowfrom ER.

Walter et al.(2012)

S-EDCL Semi-permanent

10 16* Four tripartite sensorarrays deployed,comprising of1 x 3-component sensorsurrounded by3 x 1-component sensors

Discriminated between seismicevents associated with rock falls,and slidemovement (‘slidequakes’).Slidequakes were able to belocatedwithin themoving body ofthe landslide, and the highestamplitude events associated withperiod after rainfall. Signifcantattenuation linked to surfacefissure development.

Brückl et al.(2013)

S Semi-permanent

1D 212 6* 6 seismometers Used seismometers to detectseismic events associated withmovement, and located events,including events preceding slopefailure.

Lehmann et al.(2013)

ER Controlledtest

2D 0.625 18 1 x 47m profile (48electrodes at 1m spacing

Two controlled rainfallexperiments. Compared soilwetting as measured by sesnsorsand ER measurements, andshowed wetting front evolutionthrough time-lapse images.

ER Controlledtest

2D 3 77 1 x 47m profile (48electrodes at 1m spacing

Tonnellier et al.(2013)

S-EDCL Semi-permanent

14 7* One tripartite sensor arraydeployed, comprising of1 x 3-component sensorsurrounded by6 x 1-component sensors

Firstperiod (14days) targettedsmalldisplacements, second (28 days)targettedmoderate displacements.Compared tomeasurements atValoria (see below). Located quakesassociatedwithmoving inshearingzone. Potential slow-slip seismicnoisewas identified. Low-correlationbetween increased seismic activityand landslide acceleration, butgood correlation between increasedseismic activity and heavier rainfall.

Tonnellier et al.(2013)

S-EDCL Semi-permanent

10 14* Two tripartite sensorarrays deployed,comprising of1 x 3-component sensor

Targetted large displacements,and compared to measurementsfrom Super-Sauze (see above).Identified events associated with

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 33

Table A2 (continued)

Authors

Geophysicalmonitoringmethod

Monitoringtype Dimension

Monitoringperiod(Days)

Number ofdatasets/

seismometers* Survey design Notes

surrounded by6 x 1-component sensors

material deformation. Potentialslow-slip seismic noise wasidentified. Correlated increasedseismic activity with increasedacceleration of slide.

Walter et al.(2013)

S-EDCL Semi-permanent

700 12* Three tripartite sensorarrays deployed,comprising of1 x 3-component sensorsurrounded by6 x 1-component sensors

Comparison of slidequakegeneration at three differentlandslide settings. Two of thesites (Super-Sauze andSlumgullion) reproduced datafrom previous studies, but datafrom Heumoes is from long-termpermanent array installationinstalled July 2008, following onfrom intermittent studies detailedin Walter et al. (2011). Study wasable to determine conditionsleading to brittle failure in severallandslide settings.

Walter et al.(2013)

Study describedin Gomberget al. (2011)

Walter et al.(2013)

Study describedin Walter et al.(2009)

Supper et al.(2014)

ER Semi-permanent

2D 275 1650 1 x 60m profile (61electrodes at 1m spacing

Showed time-lapse ERtomomgaphic images over periodsof varying seasonal rainfall andsnow melt.

Supper et al.(2014)

ER Semi-permanent

2D 239 1434 1 x 224m profiles; varyingelectrode spacing(1m spacing in centre,increasing at edges

Showed time-lapse ERtomographic images of amovement event.

Kremers et al.(2015)

YANGTZE-GEOproject update.

Jongmans et al.(2015)

Summary ofMainsant et al.,2012.

Gance et al.(2016)

ER Semi-permanent

2D 284 568 1 x 113m profile (93electrodes with varyingspacings of 0.5m, 1mand 2m

Used time-lapse ER tomographicimages to analyse response ofsubsurface to two natural rainfallevents, highlighting importanceof thermal exchange in ERmonitoring.

Harba andPilecki (2017)

S-ANT Transient 2D 167 12* 12 seismometers alongtwo profiles 1 x 75m and1 x 95m

Inverted ambient noise records toproduce tomographic images ofshear wave velocity. Comparativeimage analysis showed changes inslip surface and moisture content.

Xu et al. (2016) ER Transient 2D 274 6 3 x 595m profiles (120electrodes at 5m spacing

Presented time-lapse ERtomographic images to create ahydrogeological model oflandslide mass.

Imposa et al.(2017)

S-H/V Transient 2D 1589 68 34 H/V stations acrossprofiles

Compared H/V profile from alandslide 5 years apart to detectchanges in sliding surface.

Lucas et al.(2017)

ER Transient 2D 62 6 1 x 47m profile (48electrodes at 1m spacing

Used time-lapse ER tomographicimages to compare areas ofincreasing and decreasing

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 34

impact. These tend to be at least as serious as the original event, and to contribute significantly to the overallduration of the disaster’s effects.”Cohesion: In Mohr-Coulomb failure, cohesion is the component of the shear strength of a material that is notcaused by interparticle friction, and instead arises from electrostatic forces and cementing. Apparent cohe-sion can also be caused by the presence of negative pore water pressures, or soil suction.Element at Risk: Any person, property, infrastructure, or other socioeconomic unit that is exposed to a hazard.Green’s Function: Green’s functions describe the energy between pairs of seismic receivers, if one of thereceivers was an impulsive source. They therefore relate to the condition of the medium a seismic wavehas traveled through.Landslide Inventory Map: A map detailing the location, extents, and types of landslides, used for the pur-poses of hazard zoning. Landslide inventory maps may be created using a “top-down” approach, that is,

Table A2 (continued)

Authors

Geophysicalmonitoringmethod

Monitoringtype Dimension

Monitoringperiod(Days)

Number ofdatasets/

seismometers* Survey design Notes

saturation. Compared volumetricwater content measured by ERand sensors, and showed slightover estimation from ERmeasurements.

Palis et al.(2017a)

ER Semi-permanent

2D 365 365 1 x 235m profile (48electrodes at 5m spacing

Seismic monitoring focused onevent classification. Apparent ERdata clustered identified slidingmass.

S Semi-permanent

1D 365 1* 1 seismometer

Palis et al.(2017b)

ER Semi-permanent

2D 3510 3510 1 x 115m profile (24electrodes at 5m spacing

Clustered apparent ER data tolook at long-term trends ofdifferent units in response torainfall over 9.5 year period.

Provost et al.(2017)

S-EDCL Semi-permanent

40 8* Two tripartite sensorarrays deployed,comprising of1 x 3-component sensorsurrounded by6 x 1-component sensors

Tested automatic classificationmethods for detection of eventsassociated with movement.

S-EDCL Semi-permanent

21 8*

S-EDCL Semi-permanent

68 8*

Uhlemann et al.(2017)

ER Semi-permanent

3D 1369 658 5 x 147.25m profiles (32electrodes at 4.75mspacing and 9.75mbetween profiles

Presented time-lapse 3D ERtomographic images, and derivedgravimetric moisture contentfrom ER data, showing seasonalgravimetric moisture contentchanges.

Crawford andBryson, 2018

ER Transient 2D 350 5 2 x ~56.7m profiles (64electrodes at 0.91mspacing)

Presented relative changes inresistivity linked to rainfall, andresistivity results linked to shearstrength parameters; shearstrength plotted to produce shearstrength profile.

Crawford andBryson, 2018

ER Transient 2D 350 5 3 x ~62.8m profiles (69electrodes at 0.91mspacing)

Presented relative changes inresistivity linked to rainfall, andresistivity results linked to shearstrength parameters.

Merritt et al.,2018

ER Semi-permanent

1D 1740 695 4 x time-lapse contactresistance points

Contact resistances used, extractedfrom larger ER monitoring array(Uhlemann et al., 2017).

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 35

identifying all landslides within a set area, or a “bottom-up” approach, that is, identifying the hazard posed bya single landslide and identifying other similar hazards in an area.Phase Velocity: The velocity at which waves within a particular frequency envelope travel in a media, withthe principle being that different frequency waves generated from the same source may move at differentvelocities.Piezometer: An instrument for measuring pore water pressure and identifying piezometric head. The piezo-metric head represents a surface of equal water pressure, which in an unconfined system would coincidewith the top of the water table.Pore Water Pressure: The pressure of water residing in the interparticle pore spaces of a soil or rock. Porewater pressures can be negative (known as “soil suction”), in which state they tend to increase the restrainingforces against slope failure, by increasing cohesion and therefore the shear strength of a soil. Positive porewater pressures contribute to destabilizing forces in slope failure, by decreasing the effective strength of a soil.Rainfall Threshold: A value, usually in terms of amount of precipitation over time, at which failure of a land-slide is likely to occur. Increased volumes, extended periods, or a combination of these may contribute to thedetermination of a rainfall threshold.Rayleigh Wave: A type of seismic surface wave that excites a retrograde elliptical motion in the ground.Analysis of Rayleigh waves typically involves creating dispersion curves of phase velocity, instead of thefirst-arrival times used in seismic refraction surveys.Slidequake: A term by Gomberg et al. (1995) that describes seismic events generated by the motion of land-slides, rather than from sources outside of the landslide system (e.g., earthquakes, rockfalls, ambient noise).Soil Moisture Content: A measure of the water content of a mass of soil, usually expressed in terms of volu-metric moisture content (VMC) or gravimetric moisture content (GMC). Soil moisture content is determinedby measuring the volume (for VMC) or weight (for GMC) of a soil sample, before drying and reacquiringthe weight or volume to determine the change in the respective measurement.Soil Suction: The contribution to cohesive (and therefore restraining) forces generated by negative porewater pressures in a soil.Tensiometer: A type of sensor used to measure pore water pressures, including negative pore water pres-sure or soil suction.Tiltmeter: A type of sensor used to measure very subtle inclinations from the vertical that indicatedeformation.

ReferencesAcerra, C., Alguacil, G., Atakan, K., Azzara, R., Bard, P.-Y., Blarel, F., et al. (2002). Site effects assessment using ambient excitations.Al-Saigh, N. H., & Al-Dabbagh, T. H. (2010). Identification of landslide slip-surface and its shear strength: A new application for shallow seismic

refraction method. Journal of the Geological Society of India, 76(2), 175–180. https://doi.org/10.1007/s12594-010-0092-yAmitrano, D., Gaffet, S., Malet, J.-P., & Maquaire, O. (2007). Understanding mudslides through micro-seismic monitoring: The super-Sauze

(south-east French Alps) case study. Bulletin de la Societe Geologique de France, 178(2), 149–157. https://doi.org/10.2113/gssgfbull.178.2.149Angeli, M.-G., Pasuto, A., & Silvano, S. (2000). A critical review of landslide monitoring experiences. Engineering Geology, 55(3), 133–147.

https://doi.org/10.1016/S0013-7952(99)00122-2Arnone, E., Noto, L. V., Lepore, C., & Bras, R. L. (2011). Physically-based and distributed approach to analyze rainfall-triggered landslides at

watershed scale. Geomorphology, 133(3-4), 121–131. https://doi.org/10.1016/j.geomorph.2011.03.019Baroň, I., & Supper, R. (2013). Application and reliability of techniques for landslide site investigation, monitoring and early warning—

Outcomes from a questionnaire study. Natural Hazards and Earth System Sciences, 13(12), 3157–3168. https://doi.org/10.5194/nhess-13-3157-2013

Barton, N. (2007). Rock Quality, Seismic Velocity, Attenuation and Anisotropy. London UK: CRC Press.Bell, A. F. (2018). Predictability of landslide timing from quasi-periodic precursory earthquakes. Geophysical Research Letters, 45, 1860–1869.

https://doi.org/10.1002/2017GL076730Bell, R., Thiebes, B., Glade, T., Vinogradov, R., Kuhlmann, H., Schauerte, W., et al. (2008). The technical concept within the integrative

landslide early warning system (ILEWS). In Landslides and Engineered Slopes. From the Past to the Future, Two Volumes + CD-ROM. London:CRC Press.

Bensen, G. D., Ritzwoller, M. H., Barmin, M. P., Levshin, A. L., Lin, F., Moschetti, M. P., et al. (2007). Processing seismic ambient noise data toobtain reliable broad-band surface wave dispersion measurements. Geophysical Journal International, 169(3), 1239–1260. https://doi.org/10.1111/j.1365-246X.2007.03374.x

Bergamo, P., Dashwood, B., Uhlemann, S., Swift, R., Chambers, J. E., Gunn, D. A., & Donohue, S. (2016). Time-lapse monitoring of climateeffects on earthworks using surface waves. Geophysics, 81(2), EN1–EN15. https://doi.org/10.1190/geo2015-0275.1

Bièvre, G., Jongmans, D., Winiarski, T., & Zumbo, V. (2012). Application of geophysical measurements for assessing the role of fissures in waterinfiltration within a clay landslide (Trièves area, French Alps). Hydrological Processes, 26(14), 2128–2142. https://doi.org/10.1002/hyp.7986

Binley, A., Hubbard, S. S., Huisman, J. A., Revil, A., Robinson, D. A., Singha, K., & Slater, L. D. (2015). The emergence of hydrogeophysics forimproved understanding of subsurface processes over multiple scales.Water Resources Research, 51, 3837–3866. https://doi.org/10.1002/2015WR017016

AcknowledgmentsThe authors would like to thank FabioFlorindo (Editor in Chief, Reviews ofGeophysics) and six anonymousreviewers for their insightful commentsand helpful suggestions. This work wasfunded by a NERC GW4+ UK DoctoralTraining Partnership Studentship (grantNE/L002434/1) and in part by the BGSUniversity Funding Initiative (S337),which are gratefully acknowledged.Jonathan Chambers and Paul Wilkinsonpublish with the permission of theExecutive Director, British GeologicalSurvey (UKRI-NERC). No new data wereproduced in the formation of thisreview, and all data used are cited in thereference list.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 36

Boyle, A., Wilkinson, P. B., Chambers, J. E., Meldrum, P. I., Uhlemann, S., & Adler, A. (2017). Jointly reconstructing ground motion and resistivity[0.1 em] for ERT-based slope stability monitoring. Geophysical Journal International, 212, 1167–1182. https://doi.org/10.1093/gji/ggx453

Brückl, E., Brunner, F. K., Lang, E., Mertl, S., Müller, M., & Stary, U. (2013). The Gradenbach observatory—Monitoring deep-seated gravitationalslope deformation by geodetic, hydrological, and seismological methods. Landslides, 10(6), 815–829. https://doi.org/10.1007/s10346-013-0417-1

Brückl, E. & Mertl, S. (2006). Seismic monitoring of deep-seated mass movements.Carrara, A., Cardinali, M., & Guzzetti, F. (1992). Uncertainty in assessing landslide hazard and risk. ITC Journal, 2, 172–183.Carrara, A., Crosta, G., & Frattini, P. (2003). Geomorphological and historical data in assessing landslide hazard. Earth Surface Processes and

Landforms, 28(10), 1125–1142. https://doi.org/10.1002/esp.545Carrière, S. R., Jongmans, D., Chambon, G., Bièvre, G., Lanson, B., Bertello, L., et al. (2018). Rheological properties of clayey soils originating

from flow-like landslides. Landslides, 15(8), 1615–1630. https://doi.org/10.1007/s10346-018-0972-6Cascini, L. (2008). Applicability of landslide susceptibility and hazard zoning at different scales. Engineering Geology, 102(3-4), 164–177.

https://doi.org/10.1016/j.enggeo.2008.03.016Chambers, J. E., Meldrum, P. I., Wilkinson, P. B., Gunn, D., Uhlemann, S., Kuras, O., et al. (2016). Remote condition assessment of geotechnical

assets using a new low-power ERT monitoring system. 22nd European Meeting of Environmental and Engineering Geophysics, NearSurface Geoscience.

Chambers, J. E., Meldrum, P. I., Wilkinson, P. B., Ward, W., Jackson, C., Matthews, B., et al. (2015). Spatial monitoring of groundwater drawdownand rebound associated with quarry dewatering using automated time-lapse electrical resistivity tomography and distribution guidedclustering. Engineering Geology, 193, 412–420. https://doi.org/10.1016/j.enggeo.2015.05.015

Chambers, J. E., Wilkinson, P. B., Kuras, O., Ford, J. R., Gunn, D. A., Meldrum, P. I., et al. (2011). Three-dimensional geophysical anatomy of anactive landslide in Lias group mudrocks, Cleveland Basin, UK. Geomorphology, 125(4), 472–484. https://doi.org/10.1016/j.geomorph.2010.09.017

Chambers, K., Kendall, J. M., Brandsberg-Dahl, S., & Rueda, J. (2010). Testing the ability of surface arrays to monitor microseismic activity.Geophysical Prospecting, 58(5), 821–830. https://doi.org/10.1111/j.1365-2478.2010.00893.x

Chatelain, J.-L. (2004). Guidelines for the implementation of the H/V spectral ratio technique on ambient vibrations.Colangelo, G., Lapenna, V., Perrone, A., Piscitelli, S., & Telesca, L. (2006). 2D self-potential tomographies for studying groundwater flows in the

Varco d’Izzo landslide (Basilicata, southern Italy). Engineering Geology, 88(3-4), 274–286. https://doi.org/10.1016/j.enggeo.2006.09.014Craig, R. F. (2004). Craig’s Soil Mechanics (7th ed.), London; New York, Spon: Taylor & Francis.Crawford, M. M., & Bryson, L. S. (2018). Assessment of active landslides using field electrical measurements. Engineering Geology, 233,

146–159. https://doi.org/10.1016/j.enggeo.2017.11.012Cruden, D. M. & Varnes, D. J. (1996). Landslide types and processes.Daley, T. M., Solbau, R. D., Ajo-Franklin, J. B., & Benson, S. M. (2007). Continuous active-source seismic monitoring of CO2 injection in a brine

aquifer. Geophysics, 72(5), A57–A61. https://doi.org/10.1190/1.2754716Daskalakis, E., Evangelidis, C. P., Garnier, J., Melis, N. S., Papanicolaou, G., & Tsogka, C. (2016). Robust seismic velocity change estimation using

ambient noise recordings. Geophysical Journal International, 205(3), 1926–1936. https://doi.org/10.1093/gji/ggw142Deekshit, V. N., Ramesh, M. V., Indukala, P. K. & Nair, G. J. (2016) Smart geophone sensor network for effective detection of landslide induced

geophone signals. International Conference on Communication and Signal Processing (ICCSP), 6–8 April 2016 2016.1565–1569.Dou, S., Lindsey, N., Wagner, A. M., Daley, T. M., Freifeld, B., Robertson, M., et al. (2017). Distributed acoustic sensing for seismic monitoring of

the near surface: A traffic-noise interferometry case study. Scientific Reports, 7(1), 11620. https://doi.org/10.1038/s41598-017-11986-4Everett, M. E. (2013). Near-Surface Applied Geophysics. New York: Cambridge University Press. https://doi.org/10.1017/CBO9781139088435Fell, R. (1994). Landslide risk assessment and acceptable risk. Canadian Geotechnical Journal, 31(2), 261–272. https://doi.org/10.1139/t94-031Fell, R., Corominas, J., Bonnard, C., Cascini, L., Leroi, E., & Savage, W. Z. (2008). Guidelines for landslide susceptibility, hazard and risk zoning for

land use planning. Engineering Geology, 102(3-4), 85–98. https://doi.org/10.1016/j.enggeo.2008.03.022Fiorucci, M., Iannucci, R., Martino, S., & Paciello, A. (2016). Detection of nanoseismic events related to slope instabilities in the quarry district of

Coreno Ausonio (Italy). Italian Journal of Engineering Geology and Environment, 16, 13.Friedel, S., Thielen, A., & Springman, S. M. (2006). Investigation of a slope endangered by rainfall-induced landslides using 3D resistivity

tomography and geotechnical testing. Journal of Applied Geophysics, 60(2), 100–114. https://doi.org/10.1016/j.jappgeo.2006.01.001Froude, M. J., & Petley, D. N. (2018). Global fatal landslide occurrence from 2004 to 2016. Natural Hazards and Earth System Sciences, 18(8),

2161–2181. https://doi.org/10.5194/nhess-18-2161-2018Gance, J., Malet, J. P., Supper, R., Sailhac, P., Ottowitz, D., & Jochum, B. (2016). Permanent electrical resistivity measurements for monitoring

water circulation in clayey landslides. Journal of Applied Geophysics, 126, 98–115. https://doi.org/10.1016/j.jappgeo.2016.01.011Gariano, S. L., & Guzzetti, F. (2016). Landslides in a changing climate. Earth-Science Reviews, 162, 227–252. https://doi.org/10.1016/j.

earscirev.2016.08.011Gill, J. C., & Malamud, B. D. (2014). Reviewing and visualizing the interactions of natural hazards. Reviews of Geophysics, 52, 680–722. https://

doi.org/10.1002/2013RG000445Göktürkler, G., Balkaya, Ç., & Erhan, Z. (2008). Geophysical investigation of a landslide: The Altındağ landslide site, İzmir (western Turkey).

Journal of Applied Geophysics, 65(2), 84–96. https://doi.org/10.1016/j.jappgeo.2008.05.008Gomberg, J., Bodin, P., Savage, W. Z., & Jackson, M. E. (1995). Landslide faults and tectonic faults, analogs?: The Slumgullion earthflow,

Colorado. Geology, 23(1), 41–44. https://doi.org/10.1130/0091-7613(1995)023<0041:LFATFA>2.3.CO;2Gomberg, J., Schulz, W., Bodin, P., & Kean, J. (2011). Seismic and geodetic signatures of fault slip at the Slumgullion Landslide Natural

Laboratory. Journal of Geophysical Research, 116, B09404. https://doi.org/10.1029/2011JB008304Grandjean, G., Cerdan, O., Richard, G., Cousin, I., Lagacherie, P., Tabbagh, A., et al. (2010). DIGISOIL: An integrated system of data collection

technologies for mapping soil properties.Grandjean, G., Hibert, C., Mathieu, F., Garel, E., & Malet, J.-P. (2009). Monitoring water flow in a clay-shale hillslope from geophysical

data fusion based on a fuzzy logic approach. Comptes Rendus Geoscience, 341(10-11), 937–948. https://doi.org/10.1016/j.crte.2009.08.003

Grelle, G. & Guadagno, F. M. (2009). Seismic refraction methodology for groundwater level determination: “Water seismic index.” Journal ofApplied Geophysics, 68, 301–320, 3. https://doi.org/10.1016/j.jappgeo.2009.02.001

Guzzetti, F., Carrara, A., Cardinali, M., & Reichenbach, P. (1999). Landslide hazard evaluation: A review of current techniques and theirapplication in a multi-scale study, central Italy. Geomorphology, 31(1-4), 181–216. https://doi.org/10.1016/S0169-555X(99)00078-1

Guzzetti, F., Mondini, A. C., Cardinali, M., Fiorucci, F., Santangelo, M., & Chang, K.-T. (2012). Landslide inventory maps: New tools for an oldproblem. Earth-Science Reviews, 112(1-2), 42–66. https://doi.org/10.1016/j.earscirev.2012.02.001

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 37

Hack, R. (2000). Geophysics for slope stability. Surveys in Geophysics, 21(4), 423–448. https://doi.org/10.1023/A:1006797126800Haque, U., Blum, P., Da Silva, P. F., Andersen, P., Pilz, J., Chalov, S. R., et al. (2016). Fatal landslides in Europe. Landslides, 13(6), 1545–1554.

https://doi.org/10.1007/s10346-016-0689-3Harba, P., & Pilecki, Z. (2017). Assessment of time–spatial changes of shear wave velocities of flysch formation prone to mass

movements by seismic interferometry with the use of ambient noise. Landslides, 14(3), 1225–1233. https://doi.org/10.1007/s10346-016-0779-2

Helmstetter, A., & Garambois, S. (2010). Seismic monitoring of Séchilienne rockslide (French Alps): Analysis of seismic signals and theircorrelation with rainfalls. Journal of Geophysical Research, 115, F03016. https://doi.org/10.1029/2009JF001532

Hen-Jones, R. M., Hughes, P. N., Stirling, R. A., Glendinning, S., Chambers, J. E., Gunn, D. A., & Cui, Y. J. (2017). Seasonal effects on geophysical–geotechnical relationships and their implications for electrical resistivity tomography monitoring of slopes. Acta Geotechnica, 12(5),1159–1173. https://doi.org/10.1007/s11440-017-0523-7

Hibert, C., Malet, J. P., Bourrier, F., Provost, F., Berger, F., Bornemann, P., et al. (2017). Single-block rockfall dynamics inferred from seismicsignal analysis. Earth Surface Dynamics, 5(2), 283–292. https://doi.org/10.5194/esurf-5-283-2017

Hilley, G. E., Bürgmann, R., Ferretti, A., Novali, F., & Rocca, F. (2004). Dynamics of slow-moving landslides from permanent scatterer analysis.Science, 304(5679), 1952–1955. https://doi.org/10.1126/science.1098821

Hruska, J., & Hubatka, F. (2000). Landslide investigation and monitoring by a high-performance ground penetrating radar system. 8thInternational Conference on Ground Penetrating Radar, SPIE, 6.

Hungr, O., Leroueil, S., & Picarelli, L. (2014). The Varnes classification of landslide types, an update. Landslides, 11(2), 167–194. https://doi.org/10.1007/s10346-013-0436-y

Hutchinson, J. N., & Bhandari, R. K. (1971). Undrained loading, a fundamental mechanism of mudflows and other mass movements.Géotechnique, 21(4), 353–358. https://doi.org/10.1680/geot.1971.21.4.353

Imposa, S., Grassi, S., Fazio, F., Rannisi, G., & Cino, P. (2017). Geophysical surveys to study a landslide body (north-eastern Sicily). NaturalHazards, 86(S2), 327–343. https://doi.org/10.1007/s11069-016-2544-1

Intrieri, E., Gigli, G., Casagli, N., & Nadim, F. (2013). Brief communication “landslide early warning system: Toolbox and general concepts”.Natural Hazards and Earth System Sciences, 13(1), 85–90. https://doi.org/10.5194/nhess-13-85-2013

Intrieri, E., Gigli, G., Mugnai, F., Fanti, R., & Casagli, N. (2012). Design and implementation of a landslide early warning system. EngineeringGeology, 147, 124–136.

Jibson, R. W. 1993. Predicting earthquake-induced landslide displacements using Newmark’s sliding block analysis. Transportation ResearchRecord.

Jomard, H., Lebourg, T., Binet, S., Tric, E., & Hernandez, M. (2007). Characterization of an internal slope movement structure by hydrogeo-physical surveying. Terra Nova, 19(1), 48–57. https://doi.org/10.1111/j.1365-3121.2006.00712.x

Jomard, H., Lebourg, T., Guglielmi, Y., & Tric, E. (2010). Electrical imaging of sliding geometry and fluids associated with a deep seatedlandslide (La Clapière, France). Earth Surface Processes and Landforms, 35, 588–599.

Jongmans, D., Baillet, L., Larose, E., Bottelin, P., Mainsant, G., Chambon, G., & Jaboyedoff, M. (2015). Application of ambient vibration tech-niques for monitoring the triggering of rapid landslides. In G. Lollino, D. Giordan, G. Crosta, J. Corominas, R. Azzam, J. Wasowski, &N. Sciarra (Eds.), Engineering Geology for Society and Territory—Volume 2 (pp. 371–374). Cham, Switzerland: Springer InternationalPublishing. https://doi.org/10.1007/978-3-319-09057-3_57

Jongmans, D., Bièvre, G., Renalier, F., Schwartz, S., Beaurez, N., & Orengo, Y. (2009). Geophysical investigation of a large landslide inglaciolacustrine clays in the Trièves area (French Alps). Engineering Geology, 109(1-2), 45–56. https://doi.org/10.1016/j.enggeo.2008.10.005

Jongmans, D., & Garambois, S. (2007). Geophysical investigation of landslides: A review. Bulletin de la Societe Geologique de France, 178(2),101–112. https://doi.org/10.2113/gssgfbull.178.2.101

Kao, H., Kan, C.-W., Chen, R.-Y., Chang, C.-H., Rosenberger, A., Shin, T.-C., et al. (2012). Locating, monitoring, and characterizing typhoon-induced landslides with real-time seismic signals. Landslides, 9(4), 557–563. https://doi.org/10.1007/s10346-012-0322-z

Kearey, P., Brooks, M., & Hill, I. (2001). An Introduction to Geophysical Exploration. Malden, MA: Blackwell Science.Kirschbaum, D., Stanley, T., & Zhou, Y. (2015). Spatial and temporal analysis of a global landslide catalog. Geomorphology, 249, 4–15. https://

doi.org/10.1016/j.geomorph.2015.03.016Kirschbaum, D. B., Adler, R., Hong, Y., Hill, S., & Lerner-Lam, A. (2010). A global landslide catalog for hazard applications: Method, results, and

limitations. Natural Hazards, 52(3), 561–575. https://doi.org/10.1007/s11069-009-9401-4Kremers, S., Zimmermann, K., & Fleer, A. (2015). Geophysical and geodetical monitoring of slope movements at the Three Gorges

dam area of the Yangtze River in China. In G. Lollino, D. Giordan, G. Crosta, J. Corominas, R. Azzam, J. Wasowski, & N. Sciarra (Eds.),Engineering Geology for Society and Territory—Volume 2 (pp. 1415–1420). Cham, Switzerland: Springer International Publishing.https://doi.org/10.1007/978-3-319-09057-3_250

Lacroix, P., & Helmstetter, A. (2011). Location of seismic signals associated with microearthquakes and rockfalls on the Séchilienne landslide,French Alps. Bulletin of the Seismological Society of America, 101(1), 341–353. https://doi.org/10.1785/0120100110

Le Roux, O., Jongmans, D., Kasperski, J., Schwartz, S., Potherat, P., Lebrouc, V., et al. (2011). Deep geophysical investigation of the largeSéchilienne landslide (western Alps, France) and calibration with geological data. Engineering Geology, 120(1-4), 18–31. https://doi.org/10.1016/j.enggeo.2011.03.004

Lebourg, T., Binet, S., Tric, E., Jomard, H., & El Bedoui, S. (2005). Geophysical survey to estimate the 3D sliding surface and the 4D evolution ofthe water pressure on part of a deep seated landslide. Terra Nova, 17(5), 399–406. https://doi.org/10.1111/j.1365-3121.2005.00623.x

Lebourg, T., Hernandez, M., Zerathe, S., El Bedoui, S., Jomard, H., & Fresia, B. (2010). Landslides triggered factors analysed by time lapseelectrical survey and multidimensional statistical approach. Engineering Geology, 114(3-4), 238–250. https://doi.org/10.1016/j.enggeo.2010.05.001

Lecocq, T., Caudron, C., & Brenguier, F. (2014). MSNoise, a Python package for monitoring seismic velocity changes using ambient seismicnoise. Seismological Research Letters, 85(3), 715–726. https://doi.org/10.1785/0220130073

Lehmann, P., Gambazzi, F., Suski, B., Baron, L., Askarinejad, A., Springman, S. M., et al. (2013). Evolution of soil wetting patterns preceding ahydrologically induced landslide inferred from electrical resistivity survey and point measurements of volumetric water content and porewater pressure. Water Resources Research, 49, 7992–8004. https://doi.org/10.1002/2013WR014560

Leroueil, S., Vaunat, J., Picarelli, L., Locat, J., Lee, H. & Faure, R. (1996), Geotechnical characterisation of slope movements. 7 InternationalSymposium on Landslides. AA Balkema, 53–74.

Loke, M. H., Chambers, J. E., Rucker, D. F., Kuras, O., & Wilkinson, P. B. (2013). Recent developments in the direct-current geoelectrical imagingmethod. Journal of Applied Geophysics, 95, 135–156. https://doi.org/10.1016/j.jappgeo.2013.02.017

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 38

Loke, M. H., Wilkinson, P. B., Chambers, J. E., & Meldrum, P. I. (2017). Rapid inversion of data from 2D resistivity surveys with electrode dis-placements. Geophysical Prospecting, 66(3), 579–594.

Lotti, A., Saccorotti, G., Fiaschi, A., Matassoni, L., Gigli, G., Pazzi, V., & Casagli, N. (2015). Seismic monitoring of a rockslide: The Torgiovannettoquarry (central Apennines, Italy). In G. Lollino, D. Giordan, G. Crosta, J. Corominas, R. Azzam, J. Wasowski, & N. Sciarra (Eds.), EngineeringGeology for Society and Territory—Volume 2 (pp. 1537–1540). Cham, Switzerland: Springer International Publishing. https://doi.org/10.1007/978-3-319-09057-3_272

Lucas, D. R., Fankhauser, K., & Springman, S. M. (2017). Application of geotechnical and geophysical field measurements in an active alpineenvironment. Engineering Geology, 219, 32–51. https://doi.org/10.1016/j.enggeo.2016.11.018

Luongo, R., Perrone, A., Piscitelli, S., & Lapenna, V. (2012). A prototype system for time-lapse electrical resistivity tomographies. InternationalJournal of Geophysics, 2012. https://doi.org/10.1155/2012/176895

Mainsant, G., Larose, E., Brönnimann, C., Jongmans, D., Michoud, C., & Jaboyedoff, M. (2012). Ambient seismic noise monitoring of a claylandslide: Toward failure prediction. Journal of Geophysical Research, 117, F01030. https://doi.org/10.1029/2011JF002159

Malet, J. P., Van Asch, T. W. J., Van Beek, R., & Maquaire, O. (2005). Forecasting the behaviour of complex landslides with a spatially distributedhydrological model. Natural Hazards and Earth System Sciences, 5(1), 71–85. https://doi.org/10.5194/nhess-5-71-2005

Manconi, A., & Giordan, D. (2015). Landslide early warning based on failure forecast models: The example of the Mt. de La Saxe rockslide,northern Italy. Natural Hazards and Earth System Sciences, 15, 1639–1644.

Manconi, A., & Giordan, D. (2016). Landslide failure forecast in near-real-time. Geomatics, Natural Hazards and Risk, 7(2), 639–648. https://doi.org/10.1080/19475705.2014.942388

Manconi, A., Picozzi, M., Coviello, V., De Santis, F., & Elia, L. (2016). Real-time detection, location, and characterization of rockslides usingbroadband regional seismic networks. Geophysical Research Letters, 43, 6960–6967. https://doi.org/10.1002/2016GL069572

Martinez, K., Hart, J. K., Basford, P. J., Bragg, G. M., Ward, T., & Young, D. S. (2017). A geophone wireless sensor network for investigating glacierstick-slip motion. Computers & Geosciences, 105, 103–112. https://doi.org/10.1016/j.cageo.2017.05.005

Mccann, D. M., & Forster, A. (1990). Reconnaissance geophysical methods in landslide investigations. Engineering Geology, 29(1), 59–78.https://doi.org/10.1016/0013-7952(90)90082-C

Medina-Cetina, Z., & Nadim, F. (2008). Stochastic design of an early warning system. Georisk: Assessment and Management of Risk forEngineered Systems and Geohazards, 2, 223–236.

Merritt, A. J., Chambers, J. E., Murphy, W., Wilkinson, P. B., West, L. J., Uhlemann, S., et al. (2018). Landslide activation behaviour illuminated byelectrical resistance monitoring. Earth Surface Processes and Landforms, 43(6), 1321–1334.

Merritt, A. J., Chambers, J. E., Wilkinson, P. B., West, L. J., Murphy, W., Gunn, D., & Uhlemann, S. (2016). Measurement and modelling ofmoisture—Electrical resistivity relationship of fine-grained unsaturated soils and electrical anisotropy. Journal of Applied Geophysics, 124,155–165. https://doi.org/10.1016/j.jappgeo.2015.11.005

Nakamura, Y. (1989). Method for dynamic characteristics estimation of subsurface using microtremor on the ground surface. Quarterly Reportof RTRI (Railway Technical Research Institute) (Japan), 30, 25–33.

Olivier, G., Brenguier, F., De Wit, T., & Lynch, R. (2017). Monitoring the stability of tailings dam walls with ambient seismic noise. The LeadingEdge, 36, 350a1–350a6.

Öz, Y. (2015). Engineering seismology.Ozaki, Y., Mikada, H., Goto, T. N., & Takekawa, J. (2014). The 3D self-potential inversion for the estimation of hydraulic parameters. SEG

Technical Program Expanded Abstracts, 4539–4543.Palis, E., Lebourg, T., Tric, E., Malet, J.-P., & Vidal, M. (2017). Long-term monitoring of a large deep-seated landslide (La Clapiere, south-east

French Alps): Initial study. Landslides, 14(1), 155–170. https://doi.org/10.1007/s10346-016-0705-7Palis, E., Lebourg, T., Vidal, M., Levy, C., Tric, E., & Hernandez, M. (2017). Multiyear time-lapse ERT to study short- and long-term landslide

hydrological dynamics. Landslides, 14(4), 1333–1343. https://doi.org/10.1007/s10346-016-0791-6Park, C. B., & Miller, R. D. (2008). Roadside passive multichannel analysis of surface waves (MASW). Journal of Environmental and Engineering

Geophysics, 13(1), 1–11. https://doi.org/10.2113/JEEG13.1.1Parsekian, A. D., Singha, K., Minsley, B. J., Holbrook, W. S., & Slater, L. (2015). Multiscale geophysical imaging of the critical zone. Reviews of

Geophysics, 53, 1–26. https://doi.org/10.1002/2014RG000465Partsinevelos, P., Kritikakis, G., Economou, N., Agioutantis, Z., Tripolitsiotis, A., Mertikas, S., & Vafidis, A. (2016). Integration of seismic and

image data processing for rockfall monitoring and early warning along transportation networks. Natural Hazards, 83(S1), 133–153. https://doi.org/10.1007/s11069-016-2462-2

Pasquet, S., Bodet, L., Longuevergne, L., Dhemaied, A., Camerlynck, C., Rejiba, F., & Guérin, R. (2015). 2D characterization of near-surface VP/VS:Surface-wave dispersion inversion versus refraction tomography. Near Surface Geophysics, 13, 315–331.

Pennington, C., Freeborough, K., Dashwood, C., Dijkstra, T., & Lawrie, K. (2015). The National Landslide Database of Great Britain: Acquisition,communication and the role of social media. Geomorphology, 249, 44–51. https://doi.org/10.1016/j.geomorph.2015.03.013

Perrone, A., Iannuzzi, A., Lapenna, V., Lorenzo, P., Piscitelli, S., Rizzo, E., & Sdao, F. (2004). High-resolution electrical imaging of the Varco d’Izzoearthflow (southern Italy). Journal of Applied Geophysics, 56(1), 17–29. https://doi.org/10.1016/j.jappgeo.2004.03.004

Perrone, A., Lapenna, V., & Piscitelli, S. (2014). Electrical resistivity tomography technique for landslide investigation: A review. Earth-ScienceReviews, 135, 65–82. https://doi.org/10.1016/j.earscirev.2014.04.002

Pescaroli, G. & Alexander, D. (2015). A definition of cascading disasters and cascading effects: Going beyond the “toppling dominos”metaphor. Planet@ Risk, 3.

Petley, D. (2012). Global patterns of loss of life from landslides. Geology, 40(10), 927–930. https://doi.org/10.1130/G33217.1Planès, T., Mooney, M. A., Rittgers, J. B. R., Parekh, M. L., Behm, M., & Snieder, R. (2016). Time-lapse monitoring of internal erosion in earthen

dams and levees using ambient seismic noise. Géotechnique, 66(4), 301–312. https://doi.org/10.1680/jgeot.14.P.268Poli, P. (2017). Creep and slip: Seismic precursors to the Nuugaatsiaq landslide (Greenland). Geophysical Research Letters, 44, 8832–8836.

https://doi.org/10.1002/2017GL075039Provost, F., Hibert, C., & Malet, J. P. (2017). Automatic classification of endogenous landslide seismicity using the random Forest supervised

classifier. Geophysical Research Letters, 44, 113–120. https://doi.org/10.1002/2016GL070709Ramesh, M. V. (2014). Design, development, and deployment of a wireless sensor network for detection of landslides. Ad Hoc Networks,

13(part a), 2–18.Ramesh, M. V., & Rangan, V. P. (2014). Data reduction and energy sustenance in multisensor networks for landslide monitoring. IEEE Sensors

Journal, 14(5), 1555–1563. https://doi.org/10.1109/JSEN.2013.2296611Renalier, F., Bievre, G., Jongmans, D., Campillo, M., & Bard, P. Y. (2010). Clayey landslide investigations using active and passive V-S mea-

surements. In R. D. Miller, J. H. Bradford, & K. Holliger (Eds.), Advances in Near-Surface Seismology and Ground-Penetrating Radar(pp. 397–413). Washington, DC.

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 39

Renalier, F., Jongmans, D., Campillo, M., & Bard, P. Y. (2010). Shear wave velocity imaging of the Avignonet landslide (France) using ambientnoise cross correlation. Journal of Geophysical Research, 115, F03032. https://doi.org/10.1029/2009JF001538

Rosso, R., Rulli, M. C., & Vannucchi, G. (2006). A physically based model for the hydrologic control on shallow landsliding. Water ResourcesResearch, 42, W06410. https://doi.org/10.1029/2005WR004369

Safeland (2012). Deliverable 4.8: Guidelines for landslide monitoring and early warning systems in Europe—Design and required technol-ogy. Living with landslide risk in Europe: Assessment, effects of global change, and risk management strategies. Revision 1 ed.

Salvermoser, J., Hadziioannou, C., & Stähler, S. C. (2015). Structural monitoring of a highway bridge using passive noise recordings from streettraffic. The Journal of the Acoustical Society of America, 138(6), 3864–3872. https://doi.org/10.1121/1.4937765

Sastry, R. G., & Mondal, S. K. (2013). Geophysical characterization of the Salna sinking zone, Garhwal Himalaya, India. Surveys in Geophysics,34(1), 89–119. https://doi.org/10.1007/s10712-012-9206-y

Schrott, L., & Sass, O. (2008). Application of field geophysics in geomorphology: Advances and limitations exemplified by case studies.Geomorphology, 93(1-2), 55–73. https://doi.org/10.1016/j.geomorph.2006.12.024

Skempton, A., & Hutchinson, J. (1969). Stability of natural slopes and embankment foundations. Soil Mech & Fdn Eng Conf Proc/Mexico/.Spiker, E. C. & Gori, P. L. (2003). National landslide hazards mitigation strategy: A framework for loss reduction. Circular. - ed.Springman, S. M., Thielen, A., Kienzler, P., & Friedel, S. (2013). A long-term field study for the investigation of rainfall-induced landslides.

Géotechnique, 63(14), 1177–1193. https://doi.org/10.1680/geot.11.P.142Stork, A. L., Allmark, C., Curtis, A., Kendall, J. M., & White, D. J. (2018). Assessing the potential to use repeated ambient noise seismic tomo-

graphy to detect CO2 leaks: Application to the Aquistore storage site. International Journal of Greenhouse Gas Control, 71, 20–35. https://doi.org/10.1016/j.ijggc.2018.02.007

Supper, R., Ottowitz, D., Jochum, B., Kim, J. H., Römer, A., Baron, I., et al. (2014). Geoelectrical monitoring: An innovative method to supple-ment landslide surveillance and early warning. Near Surface Geophysics, 12, 133–150.

Suriñach, E., Vilajosana, I., Khazaradze, G., Biescas, B., Furdada, G., & Vilaplana, J. M. (2005). Seismic detection and characterization of land-slides and other mass movements. Natural Hazards and Earth System Sciences, 5(6), 791–798. https://doi.org/10.5194/nhess-5-791-2005

Terzaghi, K. (1943). Theoretical Soil Mechanics. London and New York: John Wiley. https://doi.org/10.1002/9780470172766Terzaghi, K., Peck, R. B., Mesri, G., & Knovel (1996). Soil Mechanics in Engineering Practice. New York: John Wiley.Toll, D. G., Lourenço, S. D. N., Mendes, J., Gallipoli, D., Evans, F. D., Augarde, C. E., et al. (2011). Soil suction monitoring for landslides and slopes.

Quarterly Journal of Engineering Geology and Hydrogeology, 44(1), 23–33. https://doi.org/10.1144/1470-9236/09-010Tonnellier, A., Helmstetter, A., Malet, J.-P., Schmittbuhl, J., Corsini, A., & Joswig, M. (2013). Seismic monitoring of soft-rock landslides: The

super-Sauze and Valoria case studies. Geophysical Journal International, 193(3), 1515–1536. https://doi.org/10.1093/gji/ggt039Travelletti, J., Sailhac, P., Malet, J. P., Grandjean, G., & Ponton, J. (2012). Hydrological response of weathered clay-shale slopes: Water infiltration

monitoring with time-lapse electrical resistivity tomography. Hydrological Processes, 26(14), 2106–2119. https://doi.org/10.1002/hyp.7983Uhlemann, S. (2018). Geoelectrical monitoring of moisture driven processes in natural and engineered slopes. ETH Zurich.Uhlemann, S., Chambers, J., Wilkinson, P., Maurer, H., Merritt, A., Meldrum, P., et al. (2017). Four-dimensional imaging of moisture dynamics

during landslide reactivation. Journal of Geophysical Research: Earth Surface, 122, 398–418. https://doi.org/10.1002/2016JF003983Uhlemann, S., Hagedorn, S., Dashwood, B., Maurer, H., Gunn, D., Dijkstra, T., & Chambers, J. (2016). Landslide characterization using P- and S-

wave seismic refraction tomography—The importance of elastic moduli. Journal of Applied Geophysics, 134, 64–76. https://doi.org/10.1016/j.jappgeo.2016.08.014

Uhlemann, S., Smith, A., Chambers, J., Dixon, N., Dijkstra, T., Haslam, E., et al. (2016). Assessment of ground-based monitoring techniquesapplied to landslide investigations. Geomorphology, 253, 438–451. https://doi.org/10.1016/j.geomorph.2015.10.027

Van Westen, C. J., Van Asch, T. W. J., & Soeters, R. (2006). Landslide hazard and risk zonation—Why is it still so difficult? Bulletin of EngineeringGeology and the Environment, 65(2), 167–184. https://doi.org/10.1007/s10064-005-0023-0

Walter, M., Arnhardt, C., & Joswig, M. (2012). Seismic monitoring of rockfalls, slide quakes, and fissure development at the super-Sauzemudslide, French Alps. Engineering Geology, 128, 12–22. https://doi.org/10.1016/j.enggeo.2011.11.002

Walter, M., Gomberg, J., Schulz, W., Bodin, P., & Joswig, M. (2013). Slidequake generation versus viscous creep at Softrock-landslides:Synopsis of three different scenarios at Slumgullion landslide, Heumoes slope, and super-Sauze mudslide slidequake generation vsviscous creep at Softrock-landslides. Journal of Environmental and Engineering Geophysics, 18(4), 269–280. https://doi.org/10.2113/JEEG18.4.269

Walter, M., & Joswig, M. (2008). Seismic monitoring of fracture processes generated by a creeping landslide in the Vorarlberg Alps. First Break,26, 131–135.

Walter, M., Niethammer, U., Rothmund, S., & Joswig, M. (2009). Joint analysis of the super-Sauze (French Alps) mudslide by nanoseismicmonitoring and UAV-based remote sensing. First Break, 27, 75–82.

Walter, M., Walser, M., & Joswig, M. (2011). Mapping rainfall-triggered slidequakes and seismic landslide-volume estimation at Heumoesslope all rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic ormechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from thepublisher. Vadose Zone Journal, 10, 487–495.

Wapenaar, K., Draganov, D., Snieder, R., Campman, X., & Verdel, A. (2010). Tutorial on seismic interferometry: Part 1—Basic principles andapplications. Geophysics, 75, 75A195–75A209.

Ward, W. O. C., Wilkinson, P. B., Chambers, J. E., Nilsson, H., Kuras, O., & Bai, L. (2016). Tracking tracer motion in a 4-D electrical resistivitytomography experiment. Water Resources Research, 52, 4078–4094. https://doi.org/10.1002/2015WR017958

Waxman, M. H., & Smits, L. (1968). Electrical conductivities in oil-bearing shaly sands. Society of Petroleum Engineers Journal, 8(02), 107–122.https://doi.org/10.2118/1863-A

Wilkinson, P., Chambers, J., Uhlemann, S., Meldrum, P., Smith, A., Dixon, N., & Loke, M. H. (2016). Reconstruction of landslide movements byinversion of 4-D electrical resistivity tomography monitoring data. Geophysical Research Letters, 43, 1166–1174. https://doi.org/10.1002/2015GL067494

Wilkinson, P. B., Chambers, J. E., Meldrum, P. I., Gunn, D. A., Ogilvy, R. D., & Kuras, O. (2010). Predicting the movements of permanentlyinstalled electrodes on an active landslide using time-lapse geoelectrical resistivity data only. Geophysical Journal International, 183(2),543–556. https://doi.org/10.1111/j.1365-246X.2010.04760.x

Wilkinson, P. B., Uhlemann, S., Chambers, J. E., Meldrum, P. I., & Loke, M. H. (2015). Development and testing of displacement inversion totrack electrode movements on 3-D electrical resistivity tomography monitoring grids. Geophysical Journal International, 200(3),1566–1581. https://doi.org/10.1093/gji/ggu483

Xu, D., Hu, X.-Y., Shan, C.-L., & Li, R.-H. (2016). Landslide monitoring in southwestern China via time-lapse electrical resistivity tomography.Applied Geophysics, 13(1), 1–12. https://doi.org/10.1007/s11770-016-0543-3

10.1029/2018RG000603Reviews of Geophysics

WHITELEY ET AL. 40